Skip to content
Demo instance Explore freely — everything resets nightly at 04:00.

What's New

See the latest updates and improvements to ProcesOS

Current Version: v3.21.0
v3.21.0

Two-step Deletions & Confirmed Certificates

Registry entries leave in two steps with a named decider, workspaces confirm each other's certificates, and more actions ask for the permission they need.

What you can do now

  • Retire a system without losing what names it Workspace

    An editor asks for an old CRM to be deleted and gives a reason. The Data Owner's bell rings, the dialog shows the processes and capabilities that still name the system, and the Data Owner archives it so those links keep their meaning.

  • Have a certificate confirmed by another workspace Compliance

    An owner records the workspace's ISO 27001 certificate and asks a partner workspace on the instance to confirm it. Once a member there with the right decides, the certificate and the ISO 27001 framework read who confirmed it until when.

  • Show confirmed certificates on your public profile Compliance

    An owner switches on "Show confirmed certificates" in the visibility settings. Visitors of the public profile see each certificate another workspace confirmed, with its standard, issuer and validity, and nothing that is only self-stated.

  • Start a team for a capability Workspace

    An admin presses "New team" on the Teams tab, picks "For a capability" and chooses one that no team takes care of yet. The capability's page opens with the team ready to be created, and nothing is created until the admin confirms.

  • See which way of running a solution fits you Marketplace

    A provider marks one Run variant for a UG or GmbH and another for a GbR. A workspace reads on the capability's page which of the two applies to its own legal form, and a team reads it for its parent's.

  • Confirm your address and carry on where you were Platform

    You sign up from a German browser and get the verification mail in German. If you open a link before confirming, a page offers to send the link again, and the link in that mail takes you to the page you wanted.

Workspace

A workspace lets a system leave the registry only through a request its Data Owner decides, and starts a team for one of its capabilities from the Teams tab.

  • Added: A registry entry leaves the CMDB in two steps: a member who may change the registry asks for its deletion with a reason, and the Data Owner archives it or declines, or the people who manage the workspace where no Data Owner is named. The requester can withdraw the request, and a captured draft is discarded in one step.
  • Added: Whoever decides a deletion, and whoever asks for one, sees what still names the entry: processes, teams below, capabilities, other entries and partner relations. An entry that nothing names can be deleted for good after its name is typed; anything else is archived and stays in the registry.
  • Added: A deletion request rings the people who decide it, waits as a row in the governance decision queue and as a next step in the Cockpit, and the system's card carries "Deletion requested". The requester hears the outcome by the bell; no e-mail is sent.
  • Added: "New team" on the Teams tab is one dialog: an empty team, a team for one of the workspace's capabilities, or a team from a template. A member without the right to manage the network creates an empty team that joins once an admin approves.
  • Changed: Deciding a capability's lifecycle gate, repairing it, changing its instances and passing a catalogue entry's review each ask the capability or curation permission; members without it see the capability without these buttons.
  • Changed: A solution made from selected artifacts and a proposed capability change take only artifacts of the workspace itself and diagrams of the workspace or of the team it created for the capability, and the capability's titles and cost figures are read from nowhere else.
  • Added: A workspace can require a second person to decide a deletion request: with the switch on the Data Owner's card in General settings, nobody decides a request they made themselves, and switching the rule off is logged and announced to the other people who decide.

Compliance

A workspace records its certificates and has another workspace confirm them, and the framework the certificate backs says who confirmed it until when.

  • Added: A workspace records the certificates it holds from the Compliance tab, with issuer, standard, number, validity, scope and the issuer's verification link and code. Each one reads "Self-stated", and the people who manage the workspace record and change them.
  • Added: A workspace asks another workspace on the instance to confirm one of its certificates, with a purpose and its consent. A member there who may verify claims decides, and the certificate then reads who confirmed it until when, with limits where they apply, or that it changed since or the confirmation ended.
  • Added: A declared framework names the certificate behind it, self-stated or confirmed by another workspace, in the Compliance tab, and an owner can switch on "Show confirmed certificates" to list confirmed ones on the workspace's public profile.
  • Changed: A system's recovery plan, its exercises, closing an incident and marking a NIS2 report as sent are for people who may change systems or assign lane stakeholders, the Head of Compliance included. Every member can still open an incident.
  • Changed: Compliance figures name their standing by meaning in neutral tones instead of red, amber and green, and the workspace export carries who may join and which body governs.
  • Changed: Compliance rule sets can read the AI facts of a step and a process, and what the registry declares about the AI system a step names, and a rule can name the day its obligation applies from, shown as "Not yet in force" until then. No rule set in this release uses them yet.
  • Fixed: The compliance views say what a rule set found in the model, with the number of rules that ran, and say when a rule set could not be evaluated. A gap's severity reads in your language, and a frozen version records whether its rule sets were fully evaluated.

Marketplace

A provider says which legal forms each way of running a solution is for, and a person sees before starting a lane quest what is still missing and how to qualify.

  • Added: A solution's Run variant says which legal forms it is for, and a capability's page says which way of running it applies to the workspace's legal form. A team reads its parent's legal form.
  • Changed: Creating, editing and rescoring a solution is for editors, owners and admins. Overriding its scores, submitting it, offering it on the marketplace and deleting it are for owners and admins, and a solution that something still names is kept until that is removed.
  • Changed: Drafting, changing and publishing a marketplace listing ask who may do that for its kind: quests for editors and the people who run them, courses for those who publish courses, everything else for owners and admins, who also withdraw a listing. Screens show only what a person may do.
  • Changed: A lane quest's page checks first whether you can start: it names what is still missing, offers "Qualify for this", and starts nothing until you press Start. A lane whose listing is archived takes no new passes, and enrolling in a course asks for its access level.
  • Changed: Blueprint search, the knowledge index and using a blueprint offer only what the asking workspace may read, and a new account's workspace starts with the onboarding blueprints the operator seeded.

People

The instance keeps a person's private goals for the companion app where no workspace reads them, and an unmet requirement leads straight to the place where it is entered.

  • Added: The instance keeps a person's private goals with their milestones and earlier versions, for the companion app to write and read: a goal can be edited, paused, retired and forked. No workspace, administration view or assistant reads them; on the web they appear as their own entry under Settings, Your data, and in the archive the person downloads.
  • Added: For the companion app, a credential says which workspace's review it rests on, when it was last checked and its status, and an unmet quest criterion names the ways the instance's records show to meet it.
  • Fixed: An unmet requirement of a personal capability links to the place where it is entered, such as Working languages or Background in the profile, and the capability's card is open again when you come back.

Processes

A diagram's versions stay with the people who may see the diagram, and a process goes to certification from somebody who may edit it or assign its roles.

  • Changed: Submitting a process for certification asks the right to edit it or to assign roles in the workspace. Everybody can still compare partners and see the estimate.
  • Changed: A diagram's versions and their figures are read by the people who may see the diagram.

Academy

A course named in a quest or a waiting room opens as the course of the workspace that names it, even where other workspaces carry one under the same name.

  • Fixed: A course named in a quest, a lane requirement or the waiting room opens the course of the workspace that names it, with its own title, also where several workspaces carry a course under the same name.

Administration

An instance admin prepares the assessment's questions and rules and publishes them as versions, while the public assessment stays closed.

  • Added: Instance admins, and anybody given the permission to manage the assessment, find "Assessment: questions and rules" under Settings, Admin, to import a rubric, check it, publish it as a version and start the next draft from it. The public assessment stays closed on every instance until its privacy wording is approved and an operator switches it on.

Platform

Anybody reads the API reference as a plain page on every instance, and a new person gets the verification mail in their own language and a way back to where they were going.

  • Added: The API reference at /api-docs is a page rendered on the server from the published OpenAPI document: who may call each operation, its parameters and a filter, readable without JavaScript and with nothing loaded from another host. The document itself is served at /api-docs/openapi.json on every instance.
  • Fixed: The verification mail and the password reset mail arrive in the language of the browser that asked for them, and a signed-in person who has not confirmed their address yet lands on a page that sends the link again and then brings them back to where they were going.
  • Fixed: On a branded instance, the legal pages and the changelog show the branded site's slim navigation and footer.
  • Fixed: Each page arrives with its theme already set by the server, and signing in applies the person's own theme without a reload.
  • Removed: The separate interactive API viewer: the API reference is the page at /api-docs.
v3.20.1

Sign-in Navigation

The navigation bar is in place the moment you sign in, and patch releases reach the changelog without an announcement e-mail.

Platform

A person who signs in lands in the app with its navigation in place, and the people on an instance get a release e-mail for feature releases only.

  • Fixed: Signing in with a password or a second factor shows the navigation bar, the What's New banner and the guided tours at once, without reloading the page
  • Changed: The release announcement e-mail goes out for feature releases; a patch release appears in the changelog and the What's New banner without a mail
v3.20.0

A Guided Start, Team Stewardship & a German-first Academy

Sign up with one password prompt and one next step, let a team take care of a capability, review a tree draft before its release, and read an Academy written in German first.

What you can do now

  • Sign up and land where you were going Platform

    You follow an invitation or a link, sign up and confirm your address from the mail. The link signs you in and takes you back to the page you came for, so you type your password once.

  • Start a workspace with one next step Workspace

    You create a workspace from a template and open its Cockpit. It names one thing to do, first modelling a process and then validating it, with one button that takes you there.

  • Let a team take care of a capability Workspace

    A workspace hands a capability to one of its teams. The team staffs the lanes from its own members and records the evidence, the parent reads it on the capability's card, and the team hands the capability back when it is done.

  • Declare the frameworks that apply to you Compliance

    An admin opens the Compliance tab and sees the frameworks suggested from the workspace's jurisdictions. They switch off what does not apply and declare a standard the workspace follows by choice. From then on the compliance figure counts exactly those.

  • Teach from courses written in German Academy

    An operator seeds the curriculum and a German-speaking team reads it as it was authored. The English follows the German with the same terms, and a reader who spots a mistake reports it on the lesson.

  • Hear about a mention by e-mail Platform

    You switch on e-mail for forum mentions and polls in your settings. When somebody mentions you or a poll you may vote in opens, a mail tells you, and you switch it off from that mail when you have had enough.

Workspace

A workspace hands a capability to a team and reads the team's evidence, starts from a template with only its own content, and sees one next step in its Cockpit.

  • Added: A capability names the team that cares for it. The team sees its parent's whole capability map, takes care of a capability from Setup and hands it back with its own management permission.
  • Added: The parent workspace sees the steward team on the capability's card and tree node and reads the team's evidence there, and the Teams tab counts what each team cares for.
  • Added: A capability's page shows its history, newest first and with the actor by name, including a hand-back and what blocked an unlock. The parent's managers are notified of a hand-back.
  • Added: The Cockpit names one next step with one primary button, and a new workspace starts with "Model your first process" and "Validate your first process".
  • Added: An appointment says which lanes stayed unstaffed and why, and the role managers are told when lanes of a newly accepted role holder were refused.
  • Added: South African Rand is available as a workspace currency.
  • Added: The Rocketlaunch package chosen on the website is preselected in the app.
  • Added: The Cockpit offers a next step only to somebody who may take it, and says when the open tasks are for the people who manage the workspace. "Validate your first process" opens the validation for that process.
  • Changed: The Cockpit's overview shows each figure once. The KPI row appears on Processes, Resources, Governance and Quests, and Manage and the Cockpit name the same biggest lever.
  • Changed: A workspace created from a template starts with its own processes, systems and roles and nothing published or staffed in its name. Review and go-live dates count from the day it was created, and a creation that fails leaves nothing behind.
  • Changed: A lane on a capability a team cares for is staffed from that team, and the lane picker offers the team's members.
  • Fixed: An accepted invitation lands in the workspace, "Open quest" opens the quest, the overview's processes open from the map widget, and course buttons open the course.
  • Fixed: Cockpit tiles and levers lead to the tab they describe, buttons name the act, and the dashboard has a direct way to the Cockpit.
  • Added: A capability gets a team of its own from its page: the team takes care of the capability and holds it as its own, and an independent organisation remains the alternative.
  • Changed: Systems in the CMDB are created and changed by editors and managers and deleted by managers and the Data Owner. Members without that right see the list and no write controls.

People

A person finds the first quest as the one next step after onboarding, and a steward's draft of a capability tree is reviewed before everyone sees it.

  • Added: A steward's draft of a capability tree passes a review round before its release: the draft carries its review state, lists what changes by name, and is requested and decided by the rule the tree sets.
  • Added: A person's capability tree shows what their household holds for cooking, baking and food storage.
  • Added: After onboarding the first quest is the one next step: the summary's main button leads to it and the dashboard shows it under the greeting.
  • Changed: The personal area is called "Dashboard" everywhere, and personal settings come in four groups: You, Security and privacy, Work and money, Help.
  • Fixed: A capability tree draft names the steward who opened it and their scope.
  • Fixed: The CV import keeps an amount that stands beside any of the product's currencies.
  • Changed: A skill claim is decided by somebody who may verify claims and is not its claimant, and workspace skills are managed by the people who manage the workspace.

Processes

A change vote that cannot open says why where the comparison is read, and an issued completion credential keeps its key through every later edit.

  • Added: A change vote that is refused while it opens in the background says why in the compare dialog.
  • Changed: A completion credential's key is frozen once it has issued: every save path keeps it, and the panel says why.
  • Changed: Data objects, decision tables, forms and resources are created and changed by editors and deleted by owners and admins. Diagram groups follow the permission for a diagram's details.

Academy

An instance reads its courses in German as they were written and in English derived from that, and a reader sees which text version they have and reports a correction on the lesson.

  • Added: Seeded Academy courses are authored in German and their English is derived from it, with a glossary that keeps the product's terms the same in both languages.
  • Added: The seeded curriculum is rewritten against the writing guidelines: the Orientation course, the workshops L0 to L7 and the role and feature courses. The handbook and the framework articles follow the same rules.
  • Added: Course translations come from a translation provider the operator configures, DeepL first, and a person confirms each one before it is published. The translations table lists every course with its coverage and names stale and hand-edited texts.
  • Added: A lesson shows the text version the instance carries, a frozen lesson is named as such, and a correction button on the lesson files one report per lesson.
  • Fixed: German texts say "Sitz" for a seat in a team, the finance quiz and the L2 workshop follow the equal split, and two lessons name who may price a change.

Compliance

A workspace declares which frameworks apply to it, and its compliance figure counts exactly those. A solution states its certificates with the issuer's verification link.

  • Added: A workspace declares the frameworks that apply to it in the Compliance tab. The suggestions come from its jurisdictions, each one can be switched on or off, and a framework without automated rules can be declared and says so.
  • Added: BSI IT-Grundschutz is in the framework catalogue as a standard a workspace in Germany can declare.
  • Added: A solution states the certificates it holds, with issuer, number, validity and a link to the issuer's verification page, on the solution, in the register and in its marketplace offer.
  • Changed: The compliance figure counts the frameworks that apply to the workspace and no others, for the workspace and for each process, and the maturity score follows it. Workspaces outside Germany see their figure move with this release.

Marketplace

A link to a listing opens that listing, however old it is, and a seeded template reads in the reader's language on every marketplace surface.

  • Changed: A seeded template reads in the reader's language in the template step and on every marketplace surface, and its chips name what the template seeds.
  • Fixed: A link to one marketplace listing opens that listing wherever it stands in the list, and a link to a listing the reader may not see says so.

Platform

A new person signs up with one password prompt and lands where they were going, and a mention or a poll reaches them by e-mail once they switch it on.

  • Added: The verification link in the sign-up mail signs the person in and brings them back to where they started, so sign-up asks for the password once. The link works once and for one hour.
  • Added: A forum mention and a poll that opens or closes can also reach a person by e-mail, once the instance's admin and the person have switched it on. A sensitive constituency is announced in neutral words, and every such mail can be switched off from the mail itself.
  • Added: Release notes carry short use cases, on the changelog page and in the release mail.
  • Fixed: A settings switch shows what the server holds when a save does not go through, and reading a thread marks its mention as read.
  • Fixed: Signing in returns the person to the page they asked for, from every page that asks for a sign-in and through the password, a sign-in provider and the magic link alike, including a settings link opened from a mail.
  • Fixed: The billing badge reflects where a subscription comes from, and the Spanish catalogue names system entries in Spanish.
  • Fixed: Orbit on the landing page answers in the page's language, and in the visitor's when they write in another.
  • Fixed: When a sign-in interrupts a Trello connection, the integrations page says that the connection was not completed and offers to start it again.
  • Added: The landing page states the FAIR certification ProcesOS holds, with the certificate number, its validity and AIFOD's verification link.
  • Added: The companion app can ask whether a quest or a course would take a person right now, and asking changes nothing.
  • Changed: A member whose membership is not yet confirmed reads and does not write, at every workspace door.
  • Fixed: The API documentation page shows the sign-in rules and API keys on every instance, and offers the interactive reference only where the instance carries one.
v3.19.0

Polls, Mandates & the Capability Tree

Poll any constituency and elect mandates from a candidate pool, grow the capability tree from the interview, turn a GitHub repository into a solution, and read a landing page that says what shipped.

Workspace

A workspace polls any constituency, elects mandates from an invited candidate pool with confirmation, invites people with grants through the waiting room, and starts communities, municipalities and households.

  • Added: A workspace opens a poll for a constituency of its choice: the governance body, all active members, the holders of a credential or a lane role. Who may vote is resolved live, a poll lives in its forum thread or on a page of its own, and the governance tab lists them all.
  • Added: An approved "found a working group" poll hands the owners a founding button that creates the team under the workspace with the poll as its record, and "New team" in the cockpit creates a team in one step.
  • Added: A cancelled poll names who cancelled it and why, and an opener withdraws an open mandate vote with a reason the voters hear once.
  • Added: A mandate carries a catalogue role and a title beside its seat type. Its candidates are the holders of a candidate credential, granted only by invitation, and the workspace chooses who elects it: the body, all active members or the holders of a credential.
  • Added: An election the body must confirm issues only after the body confirms; both stages show in setup, on the ballot and on the credential, and the mandates panel names the next election and the ballot that opened the term.
  • Added: When a term ends, the next election opens by itself with the candidate pool and the incumbent.
  • Added: An invitation or an invite link carries credential and role grants, applied once at acceptance, and invited people appear in the role, credential and candidate pickers before they accept.
  • Added: An invite link is created as direct or apply; an apply link sends the person through the waiting room and the onboarding process, and the member credential an accepted application issues names the process it came through.
  • Added: Workspace setup ends with "Waiting room and onboarding": the onboarding seats live in that step, and members come last.
  • Added: The New workspace wizard offers community and municipality beside organisation and team, and the capability interview starts a household workspace for a shared kitchen, whose members hold its resources.
  • Added: In a forum thread, @-mention a person, a group, a workspace credential or a held catalogue role, with completion in the composer; the mentioned hear the bell, a new thread rings every member who may read it, and @wizzy answers a question inside the thread, metered and stamped.
  • Added: The "Rocketlaunch" card in New diagram and the workspace wizard's review step choose a Rocketlaunch package and hand over to the billing card, and the included period ends with a 14-day notice and an opt-in: start the subscription or stay on Starter, never a silent subscription.
  • Added: Families for organisations and communities ship as released content, and a workspace's capability family reads its four words on the tree branch: status, activators, capacity, quality.
  • Changed: A proposal's vote is a poll: the hand-set "voting" status and its session id are gone, and a proposal that was stuck there returns to draft.
  • Fixed: A pending member who opens a second direct invite link gets that link's grants appended, and an apply link keeps a pending member's grants, names its refusals at the signature step and takes an accepted member straight to the workspace.
  • Fixed: The governance body saves only over what it loaded.
  • Fixed: A one-seat candidacy reads as that seat on the invitation, the link page and the queue.
  • Fixed: An unaccepted confirmation closes its election, and the election and its confirmation name each other.
  • Fixed: A vote is written with a guard, and a reader of the ballot gets no voter e-mails.
  • Fixed: A poll names its constituency by its display name.
  • Fixed: The seat-holder ladder and workspace achievements count every owner, whichever form the member row stores the organisation id in.
  • Fixed: A materialisation claim in progress is not released as stale, so two presses create one sub-workspace.

People

A person grows a capability tree from the interview, declares capabilities with evidence, always sees the next step, and a steward curates the tree in place.

  • Added: The capability tree: a person's wheel, families and definitions as a list on /journey, in the cockpit's Capabilities tab and on the node page, and on wide screens with WebGL as a three.js tree above the list.
  • Added: A person declares a capability from the tree: adopt it, add an optional claim and evidence, ask a workspace for confirmation, and see the process that uses it. Cooking, baking and food storage join the person catalogue, and the person families file their definitions.
  • Added: An empty person tree names its next step: the capability interview, then the first available twig. Adopting a capability respects the hard prerequisites the tree draws as locked and names what is missing.
  • Added: The onboarding capability interview: Wizzy proposes one item per life domain, the person confirms, and the confirmed items count as the fourth source of the first quest's step 4; a "continue" card on /journey picks the interview up.
  • Added: A tree steward curates in the tree itself: curate mode shows the draft as an overlay to add, file, move, retire and rebase, and a sector steward forks a specialisation beneath a general family, shown to the holders of that sector.
  • Added: A "mandate held" achievement appears in the badge list.
  • Fixed: The first quest's "declare what you bring" step counts confirmed background skills, not only the cockpit assessment, and its assessment step lands on the person's own surface.
  • Fixed: A capability on /journey reads "Ready to offer" only when activation would succeed; otherwise it names what is missing.
  • Fixed: A twig with no unmet prerequisite reads available, not locked.
  • Fixed: The CV import review shows a suggested skill by its name, not its ESCO identifier.
  • Fixed: The achievements route normalises conditions the way the editors do and refuses one it cannot.

Marketplace

A public GitHub repository becomes a draft solution with setup modes priced per mode and naming the partners a workspace needs, and one click records that the workspace runs it.

  • Added: "New solution from a repository" reads a public GitHub repository and creates a draft solution with its setup-mode processes; the repository's facts are read anonymously with a cache and rate limits, and capabilities are mapped from a whitelist.
  • Added: Setup modes carry their prerequisites per variant; a mode a workspace cannot deploy itself names the partners who hold what is missing, and the solution offer form prices each setup mode.
  • Added: "We run this now" creates the IT system in the CMDB from the solution's source and activates the capability nodes the workspace holds.

Academy

A course author keeps a completion credential's name editable until it issues, and the slug follows the settled name.

  • Fixed: A completion credential's slug comes from its settled name and stays editable until it has issued.

Administration

An operator runs the Rocketlaunch transition sweep as a dry run and then live, and the CV import reports when a thinking model used up its budget.

  • Added: Admin → Service purchases runs the Rocketlaunch transition sweep as a dry run first and then as a confirmed live run.
  • Changed: Qwen3.x models are treated as a thinking family, and the CV import says when reasoning used up the budget.

Platform

A visitor reads what each workspace type gets, a page for the person, the story video, dated evidence and the live count of active people and workspaces.

  • Added: Each audience page states the problems it solves, tells two use cases step by step, and lists a visitor's first steps from the persona journeys, each step badged built or planned.
  • Added: A "For you" page tells the person's side: credentials, quests, learning, founding, contributing, money, trust, the Universal Capability Economy, and POLIS as the outlook; the capability tree opens the product chapter on a three.js stage.
  • Added: The story video closes the landing page's problem chapter, self-hosted with a poster; the governance, product and pricing chapters tell what shipped since 3.15 and the Rocketlaunch packages from the price book; a social-proof section counts the people and workspaces active in the last 30 days, measured on the instance.
  • Changed: The landing pages say what the code does: the 30-day trial, on-chain minting as the Academy publisher module, the login card as an example instance, and the audience count.
  • Changed: The guided start is called Rocketlaunch in English and Despegue in Spanish, as it is Raketenstart in German; its cards in the editor, the workspace wizard and the workspace services carry the rocket, the package names on offers and invoices do not.
v3.18.0

Guided Start & Capability Families

Order a guided start in the app, file your background under capability families, open data objects from the element, and have a credential reviewed by the workspace you sent it to.

Workspace

A workspace orders a guided start in the app with a binding acceptance, a tree steward releases a capability tree version and files ungrouped codes, and data objects carry an owner and a unique name.

  • Added: A workspace asks for one of the three guided-start packages from its billing tab, states its preferred dates and whether the work is on site, and reads the offer the operator writes for it.
  • Added: The workspace accepts the offer in the app as a binding order: the terms version, the offer document's checksum and the acceptance are recorded and signed, and an offer mail names what was ordered.
  • Added: A waiting room is a door at the end of its workspace's row in the workspace switcher and on the dashboard, no longer a row of its own.
  • Added: A library data object can name an owner, a member or a process role, and every name is unique within a workspace.
  • Added: The data-object inventory opens with a way in when the library is empty, names library objects and diagram-generated objects as two kinds, and folds retention behind a disclosure.
  • Added: The capability tree is a versioned artifact: a steward releases a version, the published families follow it, and a seed click becomes a proposal the steward releases.
  • Added: A tree steward reads the UNGROUPED queue, the skill groups and occupation groups no family reads yet, and assigns each code to a family of their draft.
  • Fixed: The decision queue shows two validation sessions on one diagram as two rows.
  • Fixed: Duplicating a team diagram keeps a parent workspace's IT system without a note.
  • Fixed: A waiting room cannot be an end of any workspace connection except the one from its own workspace.
  • Fixed: A personal workspace is never offered as a claim verifier, and its settings no longer offer the claim-submission switch.
  • Fixed: A steward's saved translation of a seeded capability definition reaches readers of that language.
  • Fixed: A skill picker without a named scope moves to the whole ESCO registry when the curated set answers nothing.

People

A person files their background under capability families, has the classifier suggest one, has a credential reviewed by the workspace they sent it to, and watches a CV import run in stages.

  • Added: A person files a confirmed claim under a capability family, in the import review, in the background cards and in Settings › Profile, with "No family" always available.
  • Added: The classifier suggests a family for an imported claim from its ESCO codes, and the review preselects it.
  • Added: After a PERSON tree release, or a reapply, the classifier runs again over everyone's unconfirmed claims, so the suggestions match the released tree.
  • Added: The API answers a person's level per capability family, read on every request from their skills, stations and certificates and stored nowhere.
  • Added: A member of the workspace a credential claim was submitted to reviews the credential from the verifier queue, and a passing review raises the claim to A2.
  • Added: The person sees which workspace reviewed a credential, never the member, and a revoked review takes the A2 back.
  • Added: The CV import runs as five visible stages with a cancel, instead of a pressed button.
  • Changed: The claim-submission consent statement says that the workspace may raise the claim's assurance; older submissions can be decided but not reviewed.
  • Fixed: A field cleared on a background entry, the occupation, the issuer or the end month, is cleared on its derived assertion too.
  • Fixed: The XP at which a skill level starts is the exact inverse of the level computation, so progress bands and "Reach level n" nodes agree with the level shown.
  • Fixed: The API-key form binds a key to the chosen workspace.
  • Fixed: The CV import on a gpt-oss model gets a larger answer budget, smaller parts and both reasoning switches, and the reading says how many tokens went to reasoning.

Processes

A data object is opened, linked or added to the library at the element, a process role's rate prices a lane without one, and the cost view names a pool it does not count.

  • Added: A data object in the editor says which kind it is, library object or diagram-generated, and offers "Open in library", "Link it" or "Add to library" at the element.
  • Added: Wizzy creates a data object as a library entry, or links the one of that name, and a task's properties list the data objects it reads and writes.
  • Added: "Process Token" and "Resource Flows" in the properties panel open with a collapsed question that explains what the group is for and where its values show up again.
  • Added: A process role's default cost per hour prices a lane that has no rate of its own; a member's rate still comes first, then the lane's, then the role's.
  • Added: The walkthrough names each pool and what it is mapped to: a workspace, a type or nothing.
  • Changed: The handbook's meetings section reads the stored analysis and lists only processes with a frozen AS-IS; loading it parses no diagram.
  • Changed: Wizzy always asks before creating a data object, because the library entry outlives an undo.
  • Fixed: The cost view counts a pool mapped to a parent or child workspace as internal, and names the pool when no internal pool carries tasks instead of storing a zero.
  • Fixed: A lane rate on a pool drawn without lanes is saved and priced in every cost reader.
  • Fixed: A reader of the diagrams prices lanes by the process roles' rates, and a lane bound to a merged role is priced as the frozen version prices it.
  • Fixed: "Add to library" says why it cannot run on a diagram that is not saved yet.

Academy

A learner takes five new courses, sees why a quiz answer is right after passing, finds the way back from the Builder to the lesson, and reads where they stand on every tile.

  • Added: Five new courses: "Your background", "Meetings in ProcesOS", "Teams and workspaces", "Your capabilities and your courses" and "Admission and waiting rooms", plus lessons on the Processes tab and the capability popover.
  • Added: After a passed quiz the review of wrong answers survives a reload, and a question may carry an explanation that learners see after the pass.
  • Added: A lesson can show a screenshot with a caption and a numbered legend, and a picture without a description is refused at save.
  • Added: The Builder shows a way back to the workshop lesson a person came from, and a lesson has an address of its own.
  • Added: The orientation course stands first under "Start here", and the course page says which parts are in the reader's language.
  • Added: Course tiles say where you stand: completed, in progress with the lesson count, or not started.
  • Added: The curriculum seed brings an untouched course up to the catalogue, appending lessons and refreshing description, quiz, minutes and XP, with a preview that names every change.
  • Added: The Academy language-gap report names a missing German quiz explanation.
  • Changed: The orientation course and the L0 basics lessons are rewritten: every title is a question, nothing defends a decision, and a tripwire keeps seeded text to that rule.
  • Fixed: Course access follows the adminPanel:access permission, not the role's name.
  • Fixed: The course list is ordered before it is cut, pages honestly with a total, and the orientation course is on the first page.
  • Fixed: Quiz answers no longer stand out by length, and a pass shows the wrong answers.
  • Fixed: The course map has one scrollbar, and "All" separates articles from courses.
  • Fixed: The Spanish academy says course progress and completion in Spanish.
  • Fixed: German course texts name the skill labels the screen shows, and the orientation course has a German course record.

Compliance

An ISMS attestation shows the diagram beside its rows, and a person's export holds the decisions on their claims and their referral code.

  • Added: The ISMS attestation draws the process diagram beside the systems and data objects it asks about, and hovering or focusing a row marks its elements.
  • Added: The person's data export includes the decisions workspaces made on their submitted claims and their referral code, both listed in the personal-data register.

Integrations

An operator connects the instance to Emergence by spreetropol with a sealed token and a connection test.

  • Added: Emergence by spreetropol is connected per instance under Admin → Integrations: base URL, a sealed operator token, a connection test and a readiness word.

Administration

An operator prices and releases guided-start offers, assigns tree stewards, runs the Stripe check in the app, and reads whether an AI prompt is the current default or an edit.

  • Added: An operator prices each guided-start request in Admin → Service purchases, with the travel flat and a discount, releases the offer and later reads the acceptance evidence.
  • Added: An instance admin assigns one steward per capability tree under Admin → Taxonomy, with a handover trail.
  • Added: Admin → Founding Slots runs the Stripe check on the instance itself and shows the verdict, the unpriced modules and what each founding coupon covers.
  • Added: Each prompt in Admin → AI Settings shows whether it is the current default or an edit, and a new release's default reaches an instance that never edited it.
  • Added: The assistant's chosen name and picture reach every chat; with nothing chosen the chat shows Wizzy, and the seeded placeholder is cleared.
  • Fixed: Admin → Gamification shows a seed card only with the permission its route asks for, and a refusal names the permission.

Platform

A landing page on another domain shows the founding counter, search engines drop empty entries for private pages, and the German interface says Skill.

  • Added: The founding counter answers a listed external origin, so a landing page on another domain can show it.
  • Changed: Private areas answer every crawler with a noindex header instead of being blocked in robots.txt, so stale empty entries clear from search results.
  • Changed: The German interface calls a skill "Skill" throughout; "Fähigkeit" is the capability.
v3.17.0

Your Background & Meetings

Enter your background yourself, model standing meetings with real costs, let teams use what the workspace above runs, and read why a freeze or an import was refused.

People

A person enters their own background in the onboarding, edits it later, lets a CV prefill it, states a level from experience, and brings their own AI provider.

  • Added: Onboarding step "Your background": occupation stations, skills and degrees, entered by the person
  • Added: Your background can be edited later, in settings and on your credentials page
  • Added: A CV is read as a prefill of the three background cards, in parts that fit the model's answer budget, and every attempt says what the model did
  • Added: A self-stated skill level from stated experience, on the 10,000-hour rule
  • Added: A typed entry and the same entry from a CV are shown as likely duplicates: merge or keep both
  • Added: The person's capability cards, with /offerings on the card
  • Added: The star under the core chain, and a type chip on every card with one row per workspace membership
  • Added: A content change to a verified degree lets its verification lapse
  • Fixed: An accepted application's workspace link opens the workspace, also when its public profile is off
  • Fixed: A verification run by the subject keeps their own claim's assurance level

Processes

A process models its standing meetings with real costs, duplicates in place, and tells its validators and authors exactly why a request, a freeze or an import was refused.

  • Added: Meetings are a pickable element with their own frame on the canvas
  • Added: A meeting binds a capability, its system and a headcount per lane
  • Added: The Virtual meeting room capability and four meeting products in the catalogue, with "Add from catalogue" in the CMDB
  • Added: Wizzy creates and updates meetings with their facts
  • Added: Meeting costs: headcount, a tool share apportioned across the workspace's meetings, and a people-rate fallback, stored with the analysis
  • Added: Standing meetings as their own line in the org KPIs and the capability cost basis; a priced meeting takes the place of the tasks inside its frame wherever people cost is summed
  • Added: Meeting rows in the finance figures: people under Personnel, the room share under Systems
  • Added: The agenda per role, in the panel and inside the frame, and a meeting view opened from the panel and the context pad
  • Added: Narration and the handbook describe each meeting with its roles and agenda
  • Added: Duplicate a diagram in place, in the same or another of your workspaces
  • Added: A read-only popover on a capability reference, with six tabs
  • Added: The Processes tab lists every process, grouped by maturity, with its forum threads; the workspace figures still come from the validated ones
  • Changed: Wizzy always asks before creating a data store or a new diagram, even under "always edit"
  • Fixed: A validation request opened from the dashboard keeps its lane and quest, and never restarts the round for everybody
  • Fixed: Branch probabilities on every gateway count on a pooled process, and a note attached to a gateway is not a branch
  • Fixed: Every diagram import refusal names its reason, and a workspace export is recognised as one
  • Fixed: A contributor's own rate prices the hours that contributor carries
  • Fixed: The meeting frame's wash blends into the tasks it frames
  • Fixed: Sequence flows dock at the border of a capability reference
  • Fixed: A dropped Wizzy tool call is named where no change dialog opens
  • Fixed: A version named in a request must belong to the diagram in the path

Workspace

A team uses the systems of the workspace above it, reaches the diagram list from the dashboard, and admission runs through an onboarding body with a waiting room of its own.

  • Added: A team references the systems of the workspaces above it, grouped by their origin in every system picker
  • Added: The dashboard and the workspace shell lead to the workspace's diagram list
  • Added: A steward translates a capability definition; Wizzy drafts it, and the person's workspace pays
  • Added: Admission: application runs with pipeline progress, an onboarding body that decides applications and gets its controls in the governance setup, and the onboarding process template, seeded, adopted and bound
  • Added: The waiting room's own view for owner and deciders, and one glyph wherever a waiting room is shown
  • Added: Waiting rooms as a door on the hierarchy row, and the hierarchy folds per workspace
  • Fixed: The bakery template's skills come from the registry, and People names every lane
  • Fixed: Capability gaps are named, and only the workspace's own gaps are listed

Academy

A person takes the leadership and wallet courses, retakes a lapsed certificate, and a publisher drafts a course per lane and reads the lessons for the newest surfaces.

  • Added: A course draft per lane from the frozen AS-IS
  • Added: Certificate validity and the retake path
  • Added: Course "Agentic Intelligence Leadership" on the new leadership track
  • Added: Course "Your wallet and your credentials", ending with the wallet connected
  • Added: Lessons for the surfaces of the last two releases: teams, lane credentials, qualifying quests, module readiness, capability cards, meetings, finance reconciliation, the books import and the DPA wave
  • Fixed: Team members read a parent course's lessons
  • Fixed: The qualifying quest's title, description and bell notification are in the reader's language
  • Fixed: A lane credential naming a course is shown by the course's title in the reader's language
  • Fixed: Publishing a workspace course publishes its draft lessons
  • Fixed: The data-role lesson describes the readable "Your data" page

Compliance

A workspace sees its open ISMS review where the freeze is refused, resolves its rows by linking them, and a retired account leaves no actor id behind.

  • Fixed: The open ISMS review is named where the freeze is refused, with a link; the owner is told; unresolved systems and data objects update when they are linked
  • Fixed: Retirement tombstones the actor ids of the classified collections, and seven rows about a person get an end in a cascade
  • Changed: Nine write endpoints require an active membership

Integrations

A person connects a Cardano wallet as a personal integration, a workspace may require one, and the UTXOS wallet is offered only where the operator says so.

  • Added: A personal AI provider: your own key on your personal workspace, under Settings → Integrations
  • Added: A workspace may require a connected Cardano wallet; the wallet is a personal integration with a first-time dialog, and Admin → Integrations says whether the UTXOS wallet is offered
  • Changed: The UTXOS wallet loads its SDK from a bundle of its own

Administration

An operator seeds the capability tree, verifies the configured Stripe prices against the live account, and keeps gpt-oss models from spending the answer budget on thinking.

  • Added: The capability tree: the family registry, the curated map and the operator's seed click
  • Added: yarn stripe:verify reads the configured price ids against the live Stripe account
  • Changed: gpt-oss models are asked for low reasoning effort, so the answer budget is not spent on thinking
  • Fixed: The template list carries composedOf and every declared field

Platform

Changes are verified on the public Demo instance, and the German interface reads German throughout.

  • Changed: The TEST instance is retired; changes are verified on Demo
  • Fixed: The German interface speaks German in five more places
v3.16.0

Applications & the Waiting Room

Joining a workspace is now an application: answers, courses, evidence and a seat in the waiting room before the decision. The owner validates a lane that has nobody in it on its behalf.

Workspace

A workspace asks applicants for answers, profile fields and courses, gives them a seat in a waiting room with an orientation quest, and an Onboarding manager decides, seats them in teams and hears an appeal.

  • Added: Application policy: a workspace sets in Setup which questions, profile fields and courses it asks of an applicant, and an Onboarding manager decides beside the owner
  • Added: The request is the application: the applicant answers the questions, releases profile fields, sees the stage and the courses still open, and can sign up and apply in one go
  • Added: Application review: deciders read the answers and releases, ask for evidence, attest identity, and verify the applicant's claims
  • Added: The applicant's own applications page: open applications, evidence requests answered in place, and the workspace that asked them to apply
  • Added: The waiting room: every workspace has a hidden team where applicants hold a seat during the trial, take the courses the workspace asks for, and follow an orientation quest to a landing page
  • Added: Acceptance moves the applicant into the workspace and seats them in the teams the deciders choose; the sorting hat suggests the teams whose capabilities fit
  • Added: An appeal after a decline, once within thirty days, and the follower role a declined applicant can be offered
  • Added: The application can be signed with a Cardano wallet, and the wallet's binding is proven by that signature
  • Fixed: The applicant queue respects profile visibility, and the bell reaches every decider
  • Fixed: A viewer reads the member list and nothing more: management controls appear only with the matching permission, e-mail addresses go only to those who may invite or manage, and only an owner may give or take the owner role, at every door including setup invites and the roster import
  • Fixed: A team's contacts name the parent's designated holders until the team names its own, the team card and banner ask to join through one confirmed door, and the cockpit rebuilds its tab when you move to another workspace
  • Fixed: The two workspace exports are told apart on their buttons: "Export everything (ZIP)" and "Export the records (ZIP)"
  • Fixed: The role catalogue respects the legal form: without one, only the workspace type's default roles and the roles already held are listed, and the municipal roles carry their legal basis
  • Fixed: The bakery demo tour's "Open process" step opens the demo's main process instead of an empty editor, and a tour started without a process skips its editor steps and says so
  • Changed: Deciders of an application see only a fingerprint of the applicant's wallet address, cut on the server; the full address stays in the person's own archive

People

A person carries a membership credential of every workspace that accepted them, shows a LinkedIn link when they choose, and takes a skill off their journey without losing a workspace's verification of it.

  • Added: A membership credential is issued on acceptance and mirrored into the person's own workspace
  • Added: A LinkedIn link on the profile, hidden until the person releases it
  • Added: A skill on your journey can be withdrawn; a workspace's earlier verification of it stays logged
  • Changed: "Your data" explains every class of personal data in your language: what it is, why it is held, for how long and on what basis; code names, ticket numbers and field names stay out of the person's view
  • Fixed: The progress tour stays on the dashboard and ends beside the three doors instead of opening a blank diagram

Processes

A validation round finishes even when a lane has nobody in it: the owner signs it off on its behalf, and the frozen version says so.

  • Added: A lane with no responsible person: the owner validates it on its behalf, marked on the round and on the frozen version
  • Changed: The editor runs on bpmn-js 18.31 with the current properties panel
  • Fixed: The comparison names the ROI inputs still missing instead of showing an empty difference
  • Fixed: An AS-IS frozen by a vote, an accepted change or a seed carries its measured figures, so the comparison and the ballot show the real difference; where a side has no measurement, the dialog and the vote say so instead of showing zeros
  • Fixed: A change submitted without a price waits in "pricing open": the proposer is asked, the dashboard leads straight to the price field, and the vote opens once the price is in; a vote that cannot open says why
  • Fixed: A lane sign-off says what it achieved and how far the round is, the dashboard drops validation requests that are settled, and the XP breakdown lists every component of the total
  • Fixed: The editor header keeps its buttons clickable at every width: the state strip gives way first, "Create solution" moves into More when the row is tight, and a disabled button shows its reason
  • Fixed: A meeting's participants field renders in the properties panel, along with eleven other read-only fields that the current panel library had stopped drawing
  • Fixed: Lane credentials, task skills and resource flows saved by the editor raise no false "not valid JSON" hint, and a person seated in a lane without the credential it now requires is named in Governance with a button that sends the course as a quest
  • Fixed: When Wizzy proposes a whole process, the change dialog lists every pool, lane, step and connection it would create under the one entry, and a proposal the model could not finish writing is named in the dialog while the readable ones still go ahead
  • Fixed: The compare dialog names a lane whose people have no hourly rate among its missing inputs, whichever side built the figures, and words its cost, revenue and duration rows in the reader's language

Integrations

A person connecting Notion, Google Drive or Trello reads what they can do about it, and an operator sees every integration's setup state in one admin panel.

  • Fixed: Integration pages speak to the person: a connection that is not set up says so and points to the operator, while environment variable names, runbook paths and provider error codes live in a new Admin → Integrations panel

Marketplace

A workspace's quest list separates what is out for offers from what is assigned to a person, so a qualifying or role quest never reads as an open tender.

  • Fixed: The workspace quest list tells tenders, tasks open to members and quests assigned to one person apart: a qualifying, role or orientation quest sits under "Assigned to people" with its purpose and the person's name where you may see it, no quest shows "0 ADA", and the overview counts open quests only

Academy

The Academy reads as a path: a map by level with the next step named, one strip that says which courses workspaces are waiting for you to take, and every card saying what its certificate opens.

  • Added: Academy courses in curriculum order with plain level names; a follow-up course stays locked until the prerequisite certificate exists
  • Added: "What is asked of me" on the Academy: the courses lanes, roles, applications and orientations are waiting for you to take, the next course on the path, and the certificates you hold
  • Added: The Academy catalogue is a map: one section per level with the prerequisite arrow between them, role courses under their own heading, and each card names the application policies, lane seats and capability certifications its certificate opens
  • Added: Admin → Academy translations lists, per language, every course whose view would mix two languages, field by field, and a seed check keeps a half-translated course out of the catalogue
  • Added: "Only what is relevant to me" on the Academy: the map narrows to the courses asked of you, the next step on your path and what you have started, with the rest one click away; signed-out readers see which capability a course certifies
  • Fixed: Workspace courses have a door: the settings entry reads "Handbook & courses", a link opens the Courses tab directly, Role holders link to it, and a global course offers "Copy into my workspace" to anyone who may publish there
  • Fixed: A handbook article shows when it was last confirmed and how often it is reviewed, with the confirm button for authors, and the review interval can be set in the editor
  • Fixed: The operator's Curriculum click seeds the whole learning path, the Learn ProcesOS bundle and the workshop listings included, and the marketplace's course tab says plainly when an instance has no course listings instead of counting filters nobody set
  • Fixed: A course certificate names who issued it, the publishing workspace or the instance itself, and shows the course title in the reader's language

Platform

Support looks at an account only when the person grants it and every read is recorded, release notes arrive by area in the reader's language, and social sign-in is closed against the reported replay.

  • Added: Support access: a person grants the operator's support a read-only look at their account on request; the grant expires and every read is recorded
  • Added: The changelog and the release e-mail group each release by product area, with a sentence on what is now possible
  • Changed: Dependency updates arrive in groups without major versions; the sign-in stack and pre-1.0 packages come one per change
  • Changed: Every package the code imports is now declared in package.json instead of arriving only through another dependency; a check refuses any new undeclared import
  • Changed: The release-notes e-mail reads as a designed announcement: version and title in the header, the summary as a lead, each area with its "now possible" sentence set apart and its bullets tagged New, Changed or Fixed, and a whitelabel accent colours its button and link
  • Fixed: Social sign-in verifies the OAuth state, so a sign-in cannot be completed as another person (better-auth 1.7.7, GHSA-965c-763c-88jm)
  • Fixed: An invitation link shows the workspace, the role and who invited you, and the invitee's e-mail address is no longer part of what anyone holding the link can read or what the sign-in link carries
  • Fixed: Rendered rich text is sanitised with the current DOMPurify release
  • Fixed: An account that signs in by e-mail code confirms sensitive actions with a fresh six-digit code by e-mail, valid for ten minutes
  • Fixed: Every AI surface sends an explicit answer budget, a cut-off answer says so instead of reading as empty or half a sentence, the CV import names the cut, Admin → AI Settings warns when the chosen model spends its budget on reasoning, and Token usage labels every surface with its budget
  • Fixed: The cockpit and dashboard Wizzy say Du and tú, and offer their next steps as named links to the process in question instead of raw identifiers
  • Fixed: A first visit to the landing page opens in the browser's language, a chosen language is remembered on the next visit, and a localized address with a trailing slash lands on the page itself
  • Fixed: On an OmniRoute gateway, a model with a thinking mode is asked to answer without reasoning first, so its answer budget goes to the answer, and Admin → AI Settings says so
v3.15.0

Teams, Qualifying & Capability Cards

Teams work inside their parent workspace, a lane names the credential its seat needs, module setup runs as a quest, and every capability shows what activates it.

Added

  • Teams inside a workspace: a parent's member reads a readable team, its people, diagrams, forum and dataroom list, asks to join from the team's card or banner, and finds the teams below in the switcher and the cockpit's Teams tab; a team says how visible it is to its parent
  • Team setup: a team is born in its parent's sector and country, sets up only its profile, people, roles and decisions, adds people from its parent first with the invite link at hand, and its designations fall back to the parent's holder
  • A team works on its parent's integrations, the Trello board, the Notion share, Google Drive, Odoo with its books and the Storj bucket, unless it detaches; every gate stays at the source
  • Lane credential in the diagram: a lane names the minimum credential its seat requires, and that credential gates the seat, the execution and the quest lane; recommendations and solution views read it
  • Qualifying quest: where the lane's credential is a course, taking on the lane starts a quest to complete it, named to the person seating and to a refused runner; a course picker sets a process role's course requirement
  • Courses and certificates carry the version of the process they were written for; a new version asks for re-qualification only on the lanes that changed
  • Workspace module readiness: ProcesOS measures whether a module is configured, tested or degraded, and a capability's checklist shows the measured rows beside the human ones
  • Trello setup as a quest: a member requests activation, the admin receives a setup quest with lesson steps, the board binding and the webhook are measured, and the badge "Trello set up" follows; the course "Set up Trello" teaches the four steps
  • Capability cards in the cockpit's Capabilities tab: what activates each capability, which systems, modules and solutions, where it is used, in which processes, roles and skills, and what the evidence is
  • Capability suggestions in Setup, drawn from the workspace's own processes, systems, modules, solutions, sector and parent
  • Setup names the service that activates a capability; it becomes a self-asserted artifact of the workspace, and a solution declares which capabilities it provides
  • The meeting as a pickable diagram element with its own frame, a Virtual meeting room capability with four meeting products, and "Add from catalogue" in the CMDB
  • Personal data registry: every class of personal data states why it is held, on what basis and for how long; dataroom access requests, reward grants and credential verifications are registered; every collection is classified in code and the decision is rendered; new API keys expire within a year
  • Admin audit: every admin action leaves an audit row, and the operator can export a person's data on the controller's instruction
  • "EU processing only" keeps chat, voice and OCR on EU vendors, enforced when the setting is saved and when the call is made
  • Every workspace names its data-protection contact, and the incident runbook says who tells whom within 48 hours
  • The owner downloads everything the workspace holds, documents included, from a dialog that asks for the second factor
  • Admins, and the members of a workspace that requires it, must use a second factor after a grace period
  • One switch decides whether a member's profile is compared, a workspace can require opt-in, and you decide whether your pasted CV text is kept
  • Courses speak the reader's language and course cards show a teaser; the publishing course addresses the reader directly
  • Logs keep a shortened IP address, no secret paths and no plain e-mail addresses
  • Activator picker on a capability card: one sheet links the systems, integrations, solutions or a service by name that activate it, and one save writes the whole list; drafts and refs declared in a solution are shown read-only
  • Setup step "Capabilities", last in the hub and derived from the data: the two ProcesOS Workspace capabilities, the suggestions, and the capabilities taken in; it counts for completeness and cannot be skipped

Changed

  • The dashboard assistant receives no names, a tripwire keeps identities out of AI prompts, and a new personal workspace is named neutrally
  • Finished text extractions keep their result and drop the page text; lane seats and ids of people deleted earlier are cleared

Fixed

  • The process cost sum adds every task's cost
  • Invite links take any number of seats and days, and the join page offers both doors
  • The global bar stays in place while a session lookup is degraded
  • The login page offers e-mail and the instance's configured providers; the Google and Notion buttons are hidden
  • An archived course stays readable for the people enrolled in it, a removed course leaves its certificates in place, a course cannot enrol people into lessons above its level, renaming an article carries its courses along, and the player scrolls to the next lesson on completion
  • Admin → Members shows a last-seen date stored on the person, not read off live sessions
  • The two ?section= links in Settings open the intended panel
  • Replacing a workspace icon deletes the old image
  • The forum thread list names authors without their e-mail, no display name or personal workspace name goes out as an address, voter e-mails stay on the server, BYOK provenance names its own model, and retirement reaches extractions and cockpit notes
  • The public profile offers the ancestor join door, network candidates respect the parent's members:view, and the personal area names the person's own workspace in the bar
  • Publishing to the marketplace is open to active members only
v3.14.0

Finance Reconciliation, Role Catalogue & Continuity

The books meet the model per cost centre, roles come from a catalogue with appointments and briefs, continuity plans and incidents feed ISO 22301 and NIS2, and two courses teach it.

Added

  • Finance reconciliation: booked figures per month and cost centre, entered by hand or imported as CSV or JSON, compared with the modelled costs per category; coverage percentages for every member, amounts and hidden costs for the owner and the Head of Finance, in the cockpit's Overview tab
  • Odoo books import: the posted journal items of a month, split by analytic distribution across the linked cost centres and classified by the SKR03 or SKR04 chart
  • Role catalogue per legal form: roles with their oversight areas and legal basis, appointments as the one record, and Settings → Role holders to appoint, hand over and end with history
  • Role brief and role quest: a new holder gets a brief derived from the workspace and a quest worth 50 XP, closed when the role is handed on or ended
  • "Suggested for this role": members ranked by confirmed ESCO skills and occupation, shown to owners, admins and whoever calls a mandate vote; a suggestion only, nothing is saved by a click
  • Seven role courses, one per oversight area, recommended at appointment; a workspace may replace one with a course of its own
  • Lanes name catalogue roles, founder packages bring their role set, and an appointed holder staffs every lane that names the role
  • Academy course "Capabilities and their solutions in ProcesOS": eight lessons on defining, adopting, relating and curating capabilities and on linking, offering and adopting solutions
  • Academy course "Workspace governance in ProcesOS": nine lessons on permissions, roles, mandates, credentials, skills, occupations, badges and where they meet the processes
  • Continuity records: a continuity plan per system with a Recovery variant, exercises with four eyes and RTO copied at conduct time, incidents that set a system degraded, snapshot the cascade and start the NIS2 clock; demo seeds carry all three
  • ISO 22301 control catalogue beside ISO 42001 with the SoA switch and checks from the continuity records, the NIS2 § 30 measures folded into the NIS2 score, a five-rung BSI evidence ladder, and a lens on every framework node with OECD AI Principles and the SDGs
  • Two-factor authentication on every sign-in door, and a step-up check before sensitive account actions
  • AI provenance: before typing into the assistant, a person sees which model answers, who operates it, where, and what it may read
  • Employee data mode reaches change voting and rewards, and has its own docs page
  • Contact data leaves a CV before the AI reads it, and forum authors are pseudonymised before the AI reads a thread
  • The controller can take everything the workspace holds in one archive
  • Operator surfaces: a health endpoint with migration run state and operator alerts, retention TTLs on every short-lived collection, hosting certificates and attestations recorded with expiry warnings, self-hosted ALTCHA instead of Turnstile, Storj EU by default, and OAuth tokens, social-login secrets, invite tokens and instance AI, voice and OCR keys encrypted at rest
  • Dependabot, a dependency audit, an image scan and CSP report-only on every pull request

Changed

  • AI, voice and OCR are configured only from the database, and the vendor-side agents are gone; an instance without a stored provider and key reports the AI as off
  • Sensitive account actions ask for the second factor through the step-up flow instead of a password header; a companion app adopts the same flow

Fixed

  • SVG and PNG export draw every diagram again, including ones with capability references
  • Google Drive refuses a client id that cannot be Google's before the redirect and names the problem on the panels and in Admin → Authentication, instead of ending on Google's error page
  • Workspace deletion reaches every workspace-keyed collection, cancels the Stripe subscription first and re-authenticates the owner
  • Account deletion reaches lanes, votes, comment authors, assertions and invitations, and images leave with the account and the workspace
  • The Andamio access token is signed before it is claimed and bound to the transaction the server built, and the Web3 settings sign through the wallet without pulling the wallet library into the client bundle
  • Schema migrations carry one TTL per retention rule and every migration is recorded with its run state
v3.13.0

Trello, Offered Solutions & Phone Evidence

Trello becomes a work surface for quests and tasks, a solution can be offered and asked about, and a phone can attach photos and PDFs as evidence.

Added

  • Trello: a personal connection, one bound board per workspace, and a mirror that turns a validated process into lists and cards and a published quest into a card
  • Trello: a move onto the bound done list records the task's execution when the lane names one responsible person
  • Solutions are offered in the marketplace with setup, retirement and per-model prices in euro; buying is an inquiry the provider answers or closes
  • "Does this fit us?" compares a run model's required skills and people against the buyer's workspace and names what is covered, partial or missing
  • A solution derives its audience from the pool in which a capability is activated, and states the capacity the target's role profile holds
  • Photo and PDF evidence files per listing, readable only by the listing's participants and attesters; references are validated
  • Personal quest views and passes as API for the companion app; quest notifications carry the listing id; listings can be fetched by slug
  • Occupations in a solution's derived facts show their names, and a system operated elsewhere names that workspace
  • A mirrored Trello card pasted as an evidence reference shows as a card with its name
  • Employer hourly rates per person, kept with their history and readable by the owner, the head of finance and the person themselves; a lane without its own rate reports the average of its assigned people and names who has none

Changed

  • A solution inquiry can be marked answered or closed by the offering workspace's owners and admins, and the asker is told
  • A team workspace requires a parent workspace: both creation paths refuse a team without one, and the workspace wizard asks for the parent before it can finish
  • Deleting a workspace removes its evidence files, its Trello binding, mirror rows and webhook deliveries, and deregisters the board's webhook; retiring a person anonymises their uploads, and both data exports list evidence files as metadata
v3.12.1

Course Player and Recalculate Fixes

A course with every lesson read no longer reloads without end, the reader offers the next lesson, and Recalculate in the editor stores the figure it computes.

Added

  • The lesson reader offers the next lesson, and only "Back" on the last one

Fixed

  • A course with every lesson read and a practical component still open renders once and names the blocker instead of reloading without end
  • Recalculate in the editor's status strip stores the monthly cost it computes, on frozen AS-IS versions as well, and the Versions panel shows the same figure
v3.12.0

Solutions Read Their Diagrams

A solution now takes its capabilities, skills, systems, duration and audience from the diagrams it links, and a workspace blueprint in the marketplace becomes a workspace with one click.

Added

  • Solutions derive capabilities, skills, occupations, systems with dependencies, duration, throughput and audience from their linked diagrams, per run variant split into provider and customer side
  • The solution form shows "From the process" beside "Added by you", locks derived audience types with their origin, flags additions that no linked diagram carries, and offers to adopt what the diagrams name
  • A pool without a workspace names one of the fourteen workspace types and, optionally, the capability whose team stands in it
  • A workspace blueprint in the marketplace is created as a new workspace, a team or a sub-workspace under a workspace you administer, and the workspace wizard accepts a preselected blueprint
  • "Learn ProcesOS" is a workspace blueprint: creating it gives a workspace with the eight workshop capabilities ready to be held

Changed

  • The three legacy pool categories are migrated to workspace types (consumer to person, business to organisation, government to municipality)
  • "Adopt into workspace" no longer appears on workspace blueprints, since it recorded an adoption without creating anything; the create action replaces it
  • The Continuity & Resilience product id is part of the shipped product map, so a dedicated instance needs the environment override only for its own Stripe account
v3.11.0

Continuity, Execution Governance & Passes

Ask what stands still when a system fails, keep agents inside their mandate at the moment of acting, and let every helper run their own pass of a process.

Added

  • Continuity & Resilience module: a per-system failure cascade over dependent systems, tasks, processes, capabilities and partner workspaces, in the cockpit Systems tab, the editor's analytics and Wizzy's stand sentence
  • Typed system dependencies (required or optional), stand-in systems, and the workspace that operates a system; a shared service names the systems it shares in both directions
  • Recovery targets as numbers on processes, systems and capabilities, with the cascade naming which system comes back later than a process can wait
  • Capabilities show a derived degraded state with the systems that cause it; the execution gate reports it to persons and refuses agents
  • Execution events: one append-only record per lane execution with a hashed input and the mandate it ran under, a pre-action gate, handover entries per pool boundary, and an oversight register in the cockpit
  • Passes: each participant runs a lane on their own, delivers rounds at the diagram's gate, is paid per confirmed round, and finds their passes in one list
  • A change process with a plan is executed as quests; the last lane's confirmation re-freezes the process as its next AS-IS
  • Handbook articles carry a review interval with a one-click reconfirmation; the cockpit lists the ones past it
  • A personal process can be confirmed lane by lane as its own status, distinct from validation
  • An accepted interview proposal rings the process owner, and an accepted "there is no system" converts the task on the server
  • Demo workspaces carry criticality, single points of failure, dependencies and recovery targets, so the cascade has something to show

Changed

  • The AI workspace briefing names special-category data objects by class only, and the provenance answer says what was withheld
  • Defining XP achievements and skill trees requires the workspace-management permission
  • The session list identifies devices by an opaque handle

Fixed

  • Lane names with special characters register under their real name in both lane parsers, and the credential gate for such a lane resolves its role
  • The federated public workspace list returns workspaces that share their location
  • A locked thread shows its status to the people who may see it
  • Template export carries system dependencies and integration endpoints as matching placeholders, without secret references
  • Editor documentation reflects the document import that shipped, and the billing designation says why it grants nothing
v3.10.0

Curriculum, Workspace Courses & Priced Change

A course for every maturity level, courses your workspace publishes itself, and a change decision that shows what it costs and when it pays off.

Added

  • Curriculum courses L0 to L5 with workshop diagrams, quizzes and a practice component that measures the learner's own diagram; L6 and L7 as an outlook
  • Practice mode in the validation wizard for rehearsing a sign-off on a course diagram
  • Operator seed for the whole curriculum, one click or one CLI command, and the "Learn ProcesOS" marketplace bundle of the eight workshop capabilities
  • Workspace courses: create, adopt from the catalogue and adapt, publish to the teams below, sell as a listing, or propose as a standard for the instance
  • Course completion credentials issued by the publishing workspace, and a course badge that names it
  • A course for Academy publishers, nine lessons with quiz
  • Change process per AS-IS → SHOULD-BE pair with a one-time change cost, a cached comparison, and the amortisation date
  • Decision surfaces show the same four figures: change cost, monthly cost before and after, and when the change pays for itself
  • Wizzy's review of a diagram as evidence on a quest attestation, and as feedback on a course practice diagram
  • Solution variant pickers offer agreed AS-IS processes, and submitting a variant requires one
  • Interview answers are reviewed by the process owner; accepting one writes it into the diagram, rejecting one records the reason
  • Data repairs in the admin panel: recompute maturity levels and remove forged AS-IS rows, each behind a dry run
  • API keys can be scoped to a workspace from the form
  • A first-time guest is greeted in the language their browser asks for

Changed

  • A change vote opens only once the fork is priced; an unpriced fork waits in "pricing open" and its creator is asked
  • The change cost is the whole price of a change; the earn-as-you-grow implementer share is retired
  • The L5 gate requires a priced and compared change; the L4 and L5 lessons and the docs say so
  • Learning capabilities are excluded from every maturity denominator
  • Members leave a workspace through ProcesOS's own exit; the framework's route is closed
  • The editor's state strip and analytics dialog show the level the server computed and its gaps
  • A membership still waiting for the governance body can read, and nothing else
  • Seeded template processes are validated through the real validation path, or stay drafts

Fixed

  • The validation server derives lane, tasks and quests from the diagram, so a session signs only what its lane owns
  • Sign-off snapshots merge per lane and detect in-place task edits; every save is compared with the snapshot
  • Every branch of a decision gateway asks for a probability in all four places, and inclusive gateways may total more than 100%
  • Manual AS-IS freezes ask "validated first", and the maturity ladder computes L4 to L7 live from the stored diagram
  • A marketplace bundle lists what is inside it, a prerequisite shows its title, and a quest states what it pays and in what rhythm
  • Every workspace permission is named and explained in the roles editor, and the legend beneath speaks the admin's language
  • Screen readers hear labels in the user's language; the Spanish settings page is in Spanish
  • Timespans are priced down to the second and the panel names each one
  • The wallet portfolio reports a missing network instead of failing
  • The delegated FAIR registrar reaches the admin shell
  • The acceptance progress card is reachable and knows all eight gates
  • Quiz answers appear at varied positions and shuffle per attempt
  • A markdown link onto an app route keeps its destination

Removed

  • Implementer share slider and the client-net-savings split in the ROI view
v3.9.1

Migration Hotfix

Schema migrations create the collection they index, so the 3.9.0 upgrade applies cleanly on every instance.

Fixed

  • Schema migrations create the collection they index when it does not exist yet, so an upgrade records every migration and the instance no longer reports itself degraded
  • The same guard reaches the foundation-skill slug index, so instances seeded by earlier tooling apply that migration cleanly too
v3.9.0

Wizzy in the Cockpit & Credly Badges

Wizzy reads your cockpit and speaks up, credentials become Credly badges, a process can start from a document or a conversation, and every template ships as a spec.

Added

  • Wizzy on every cockpit route: reads the areas you may see, names your standing, suggests only resolvable actions, and is metered and stamped
  • Cockpit notice: one computed sentence about what needs attention next, dismissable and never repeated
  • Credly integration at instance level: badge designs per credential kind, opt-in per person, mirror button, automatic issuance per design, revocation sync
  • Quest completions carry a stable badge kind and the quest builder can attach a Credly badge design
  • Create a process from a document: text layer first, OCR fallback, candidates with page citations, review, import through the pack importer
  • Process conversation: tell Wizzy a process, it asks along the workflow; "Analyze with Wizzy" builds the diagram with a source turn per element
  • A capability node shows which exchange solutions cover it, grouped by what each solution still needs
  • The Gaps tab lists real gaps and names why a dependency is missing
  • A skill certification can be issued by hand by the Head of Education
  • A published capability can be taken into an organisation's tree from the exchange
  • A workspace states its LEI; the register is asked and re-asked, and a DNS record binds the identity
  • The discussion of a process is reachable from the editor, with its message count
  • A solution listing has an address on the exchange, and register slugs link to the capability node

Changed

  • Feature tiers now bind server-side: a route refuses a workspace below the feature's tier, not only one without the module
  • Every workspace template, the Village, the demo and the blueprint scripts are built from process specs; no seed carries hand-written BPMN any more
  • Seeded lanes carry a rate with a time unit, start events carry tokens, meetings are groups — the figures stop reading as zero
  • Achievement suggestions come through a metered, stamped route and the rules table prices them; the AI skill-suggestion button is retired
  • One way to a new diagram, in the editor and on the overview; the last edit before "+ New" is saved
  • The Academy shows a plain sentence and the handbook links while the catalogue is empty; filters appear once there is something to filter
  • Governance and Versions tabs say why a button is unavailable, and stop offering what has already happened
  • The workspace creation wizard names its steps consistently, describes every module, counts parent workspaces and marks the irreversible choice
  • Meeting captions render in the reader's language from the stored facts
  • Reward grants and approval requests lead to the Rewards tab

Fixed

  • A credential revoke reaches only the workspace the caller is allowed to act in
  • The level-reward list answers members of the workspace only
  • A personal workspace admits nobody but its owner
  • Head-role designations and the workspace currency are read from the organisation row again, so designated heads count and cost bases show the set currency
  • A lane role's skill requirement is a minimum, and a minted certificate meets it
  • A workspace skill without an active flag counts as active for every reader
  • The validation round hands out its tasks again when a single person holds a lane
  • The structure delta and the walkthrough handle ad-hoc sub-processes on one level
  • The write path repairs capitalised BPMN tags and records what it repaired
  • A person's credential list loads on an instance without a chain
  • Seven properties-panel entries reach the screen, and one throwing entry no longer freezes the panel
  • The Soziale-Fahrten and Supply-Chain templates find their data objects in the library
  • The onboarding, the finish screen and the dashboard agree on what a complete profile is
  • A non-member who follows a shared workspace link lands on the public profile or the dashboard instead of a dead end
  • The instance node reaches active where the earlier migration could not
v3.8.0

Companion App and Personal Integrations

The phone signs in and gets notified, Notion and Google Drive connect to the person rather than the workspace, and a role offering becomes an engagement that pays out of what it improves.

Added

  • The companion app signs in with a bearer token issued from a browser session, and workspace API keys are refused on personal endpoints
  • A push channel to iOS and Android for the companion app, carrying the notification's group and its id and nothing else; an instance sends once the Apple and Firebase credentials are placed
  • Notion connected per person: pages as workspace knowledge, databases mapped onto resources with a dry run, pages and databases as dataroom sources
  • A Notion database bound to a start event counts the runs of a process, so the frequency behind every cost figure is measured instead of estimated
  • Google Drive connected per person, with a personal dataroom that can be bound into a workspace
  • Workspace members can be invited to a bound Drive folder through Google, and the workspace role now decides the Drive role
  • A role offering becomes a capability engagement, with the outcome share credited from the measured KPI difference
  • Quest bounties are recorded as payout credits in a ledger, with per-currency totals and a stated gap where no payment rail exists yet
  • The founder pack as a one-time offer: a consultation and a legal-form template pack delivered into the personal workspace
  • A workspace exports its own records as an archive, with a manifest naming what it contains and what it deliberately does not
  • Solution packs import as a zip with a manifest, a dry run first, a cost estimate and a funding request to the head of finance
  • Solutions carry the ESCO skills their lifecycle processes demand, the data flows between their systems, and the parts each cost figure is made of
  • A start event can carry a batch size, so a run that moves twenty-four items costs twenty-four items
  • The quest pane says what a quest is about, whether that is a capability, a solution or a resource
  • Odoo Documents folders as a dataroom source
  • Wizzy can create a data object and a data store, the store linked to the workspace CMDB, and can insert a step between two others by making room instead of hunting for it
  • A meeting drawn on a diagram carries its participants, duration and cadence, and is costed from the attendance of every lane in the room

Changed

  • Rearrange and Tidy up lay a diagram out on the same grid the spec builder uses, instead of a second arrangement
  • The eight collections that hold personal data have a class in the register and a step in the retirement cascade
  • Setup steps report their completion from the data behind them, and the systems step exists rather than counting as a placeholder
  • The register of the FAIR passport has an address, and three surfaces lead to it
  • Odoo handbook articles go through the same sanitizer as everything else, with one configuration instead of two
  • Seeded sample workspaces and the process-lifecycle handbook article read in the language the interface is set to
  • The six leadership roles are named in German on a German interface, each with a sentence saying what it does and whom it reaches, so the billing contact and the head of finance are told apart
  • The help panel lists the knowledge base and each workspace's handbook, and the global menu has an entry for it

Fixed

  • A re-keyed lane assignment keeps the workspace role it was vouched for, and the person saving the diagram is told when a rename could not be carried over
  • The account deletion reports what it removed and what it kept without a name, in sentences a person can read
  • Forking a frozen diagram asks the same credential gate the direct assignment asks
  • The cost view names a per-execution figure as such once a run carries more than one token
  • The editor toolbar fits its window: controls fold into a disclosure as the window narrows, so the title, the tabs and the buttons stay readable and separate from about 930 px upwards
  • The header names the workspace you are in, including one you just created
  • The closed state strip announces the frozen lock and the maturity level to a screen reader, not only to the eye
  • Opening the editor no longer leaves aborted view transitions as unhandled rejections in the console
v3.7.1

Walkthrough Fixes II

The German interface speaks with Du and one word for the workspace, the first screen after sign-in speaks once, and the remaining findings of the walkthrough are closed.

Added

  • Post-sign-in landing on the dashboard with one address at a time: cookie notice, tour offer, release banner
  • Leadership roles default to the workspace owner at creation; existing one-person workspaces are backfilled once
  • A FAIR register entry in the user menu and the landing footer
  • Quest listings show bounty, goal, deadline and the issuing workspace

Changed

  • German interface: Du everywhere except legal, invoice and privacy texts; one word for the workspace
  • Template figures carry a stamp until the first own change and are excluded from maturity; a seeded vote runs as the template
  • The maturity card names its profile and shows a medal only when more than half of the profile's dimensions are assessed
  • The head bar reads XP and notifications from one store per tab
  • Framework names follow the reader's language on every surface; the data-object register lists each open item once
  • The compass appears above a threshold of public workspaces and marks the one list; its reasons name what matches
  • Wizard and setup copy say what happens: Continue buttons, box placement, the irreversible choice named, entry counts measured, the clarification boxes shortened
  • Dates and numbers follow the reader's locale on the offerings page and invite links
  • The tour announces its continuation instead of starting it, every step highlights something or stands alone without darkening the canvas, a click beside Wizzy asks before it ends the tour, and the bubble waits for its anchor

Fixed

  • The working-languages field keeps what you type
  • Avatar initials skip brackets and digits
  • The sixth leadership role can be assigned in the workspace settings
  • Process tiles in the cockpit open the process and are reachable by keyboard
  • A quest without requirements no longer certifies that you meet them
  • Seeded data objects of the bakery template resolve to the library, so a new workspace starts without an Art. 30 gap
  • Level-gate promises that nothing enforced are gone from the product and the code
v3.7.0

Spec Builder & Datarooms

Every full page load renders again, a process is built from one spec, Notion and Odoo Documents mirror into datarooms, and forty defects from a full UI walkthrough are fixed.

Added

  • Spec builder as an application module: one call builds a laid-out diagram with readiness findings and a solution pack; the village template seeds through it
  • Wizzy builds a process from a spec (`build_process_from_spec`) behind the usual approval, and relays import warnings and pool or skill gaps
  • Diagram list imports several `.bpmn` files at once, each with an optional `.pack.json` that creates systems, data objects and token resources; ignored files are named
  • Wizzy points at the element he talks about, asks before walking a token, and looks at the diagram as a picture on an instance whose model can see
  • Notion: personal connection, sharing of a page or database into a workspace, and a search across the connected space
  • Datarooms mirror a Notion page or database or an Odoo Documents folder as metadata, with sync runs, a run history and a mirrored-entry count
  • Odoo Documents: the integration browses folders and files on Odoo 18 and reports whether the Documents module is present
  • Dashboard: three doors after the Core activation, the chain from the instance to you to your workspace as one picture, and the first quest one scroll away
  • Tour: the start names its reason, the wait is narrated, a locked tour names what would unlock it, and a skipped tour is never recorded as taken
  • Release images carry a version tag and OCI labels; the main branch publishes its own tag

Changed

  • Templates seed drafts with their quantities; they no longer seed validations, task confirmations or frozen AS-IS versions in the creating person's name
  • One process count on every cockpit tile, sentence and map; a SHOULD-BE fork is not a second process, and the two validation terms are named apart
  • Compliance rulebook explanations for LkSG, NIS2, GDPR, DORA and ISO 27001 read in the reader's language, with plurals by number
  • Numbers, currencies and relative times follow the reader's locale everywhere; the save pill and a language name speak the reader's language
  • Maturity exclusions read as sentences instead of internal codes, and capability nodes show their titles in governance blocks
  • AI readiness is answered once, for the About-Wizzy panel, the readiness bubble, the presentation and the interview; a workspace without a provider gets a sentence and a reference, never a bare error
  • The editor derives its workspace from the loaded diagram, the versions list follows the diagram, and Manage diagrams returns to that workspace's list
  • The billing overview shows the instance licence and no upgrade hint while the licence holds
  • The treasury lock distinguishes a role held by someone else from a role nobody holds, and links the step that assigns it
  • The personal onboarding screens say what to do, the Settings shell scrolls once, and locked tabs read as locked
  • The instance card on the journey page, the settings menu entries for data objects and artefacts, the tour's step about levels, and the Escape key in the workspace switcher behave as their words say
  • Admin, Foundation Skill Library: the sector tiles count the whole registry, the occupations are listed by ISCO major group, and the delete dialog opens as a dialog
  • The governance margin is profit over revenue, averaged over the processes that have revenue, and says over how many

Fixed

  • Every full page load with a session renders: refresh, Back button, bookmarks, links from email, the reload after a language or theme switch, and the 404 page
  • A federation identity that does not exist is an empty state, not an error
  • Failed requests carry a reference the person can quote: credentials, diagram import, presentation
  • Avatar initials skip brackets and symbols on every avatar
v3.6.1

Interview Voice Patch

Wizzy speaks one thing at a time in the interview, offers one dialog per system, and recognises a system named with words around it.

Changed

  • A mention that carries a known system name as whole words links to that system rather than offering a new one

Fixed

  • Wizzy reads the task sentence and the interview question as one utterance, and a newer utterance replaces one still being synthesised
  • A system named in the answer opens one confirmation dialog, and a name declined once is not asked again during the session
v3.6.0

Wizzy, the FAIR Register and AI Provenance

Wizzy guides the tour and drafts quests, the FAIR Passport gets a public register and a registrar desk, and every AI-written output says so.

Added

  • Wizzy guides the product tour as an animated figure with a pointing pose and a speech bubble; the tour is step one of the first quest "Wizzy shows you ProcesOS"
  • The editor walkthrough highlights, scrolls to and selects the current element on the canvas
  • A public register of FAIR Passports with facets, search and sort
  • The registrar worklist as an admin panel, with an application detail showing the trail, moves, passport status, revocation and the assessment lane
  • A case lane on the application detail: fee waiver, FFDE accompaniment and appeals, each with its refusal spelled out
  • The instance is a real, active node in every workspace tree, with a card on the journey naming the AI model and provider, the operator, the region, the licence, the federation identity and the data-centre certificates the operator enters
  • Wizzy presents the walkthrough: the figure moves to each element with the narration in its bubble, the interview question answered in place, and the text panel as the fallback for reduced motion and small screens
  • Framework agreements between a person and the operator or a workspace, on versioned templates per jurisdiction, with terms acceptance as the floor and money exchange locked until the agreement is active
  • A provenance stamp and badge on every AI-generated output: diagrams, mirrored comments, forum insights, process drafts, reports, quest drafts and Solution Package sections
  • Voice transcriptions name their provider and processing region
  • A five-step quest builder from the need to the publish, Wizzy drafting a quest from free text, and eleven partner standard quests one click away
  • Workspace type, solution portfolio and FAIR Passport as quest criteria; solution-targeted quests are suggested to the workspaces that carry the solution
  • The fulfiller's view of quests, with a balance that follows the scope and names the income nothing records
  • A frozen AS-IS sends you to a fork, and an accepted change vote makes it the next AS-IS
  • A validation quest per lane with a stepper in token-flow order, and what changed in the process on the page people vote from
  • One ballot with several candidates for one seat
  • A network template with a council, three sub-groups and one invite link
  • Federation peers by URL with a real handshake from the panel and incoming requests
  • Location once, as a jurisdiction code, with one classifier for the six EU sites
  • Four self-asserted person status artefacts as the placeholder for the EUDI wallet
  • Kind and exchange on personal offerings; legal standing as an activation requirement asked only when money is exchanged
  • Compliance shows the AI system per task and the governance owner's scope
  • A Forks filter in the diagram listing, off by default
  • Skills by life domain after a CV import, the profession step, and a way out of the data-room wall on the journey
  • A cacheable prompt prefix with a workspace briefing and a TOON delta, and a "From cache" column in token usage
  • The wizard prompt names every tool, and the TOON legend is generated from the code
  • OmniRoute as a speech provider with seconds from the gateway, a model-name capability map with grouped selects, OCR model suggestions, and the effective AI source with usage grouped by model and area on the admin page
  • ESCO 1.2.1 as the pinned skills release, with an operator runbook

Changed

  • Standard quests are the one path to a partner; the partner-request API is retired
  • The provider registry carries its own freshness check; the public CMDB scan without a caller is gone
  • A published listing keeps its contract terms; changes go through a new listing
  • A diagram's governance status moves only through the governance flow, never through a metadata edit
  • Internal quests are visible only to members of the publishing workspace
  • Wizard readiness counts confirmed members only

Fixed

  • The properties panel, the element type in its header, the save summary, the GTT capability shape and the Sustainable Impact evidence chips speak the reader's language
  • Structured AI answers arrive complete on gateways that stream by default
  • A governance promotion interrupted mid-way completes on the next run
  • Quest deadlines that arrive as text are stored as dates
  • The template picker asks the registry which sector a template is for
  • The admin persona travels with the wizard prompt, and the admin page shows which prompts are overridden
  • One compliance vocabulary across the editor tool, the panel and the maturity gates
  • Task and process compliance are recorded in the maturity snapshot
  • Icon buttons on the FAIR surfaces carry names for assistive technology
  • The Spanish FAIR catalogue is fully translated
v3.5.0

Confirmed Members

A governing body confirms who joins and issues a member credential, the operator can invoice and sell AI token packages, and ESCO skills reach the picker, solutions and life domains.

Added

  • Membership confirmation: a workspace with a governing body holds joins from links and join requests until a chair, vice-chair or operator confirms or declines them
  • Pending members read everything and change nothing, see a banner saying so, and receive a member credential from the body on confirmation
  • Payment requests with a Mod 97-10 checked reference, settled only by the ledger or the Stripe webhook, with part payment and overpayment named
  • AI token packages a workspace can buy once the operator configures them, credited as an addition to the monthly budget for the months bought
  • Fulfilment invoices behind the merchant-of-record switch, a printable invoice document, and an invoicing settings page for the operator
  • ESCO fetch as an admin action in the background, with progress, a job that survives a page reload, and a notice when a newer ESCO release is known
  • Skill groups and education fields as an axis on foundation skills, with labels in every shipped language, and group headings in the skill picker
  • Directed skill relations from ESCO, so a skill that is essential for another counts as its prerequisite
  • Workspace skills can be linked to a foundation skill, and linked skills satisfy capability requirements and quest criteria that name the ESCO skill
  • A solution's skill and occupation profile, derived from the capabilities it names, with rows the provider adds
  • ESCO skill groups, education fields and occupation groups mapped to the seven life domains as anchors, shown in the life-domain admin panel
  • Operator runbook for fetching, mounting, seeding and upgrading the ESCO dataset

Changed

  • A capability requirement on a skill nobody in the workspace can be evaluated against reads as not evaluable, with the reason named, rather than as unmet
  • The seed button in the life-domain panel is always offered, and the seed reports which anchors it added
  • The skill detail panel and the skill graph show the four ESCO relation types with their own labels

Fixed

  • The admin AI top-up grant is credited once per payment request, however many bank lines settle it
  • The ledger match form offers payment requests, so a short transfer marks the request part paid
v3.4.0

Quests From Process

A validated Setup process becomes a quest graph, a workspace elects its governing body with term limits, courses end in a certificate and a badge, and the operator's bank account is wired in.

Added

  • Quest from process: a validated Setup diagram generates a quest graph, one quest per lane, with dependencies, pre-assignment and credential gates
  • Six quest kinds (setup, run, change, retire, work, delivery), each with its own acceptance rule, and a typed target of capability, solution or resource
  • Recurring quests: run services and deliveries run in cycles that are delivered, confirmed or rejected, with the amount per cycle declared up front
  • Quest roll-up: when every lane quest is verified the capability is ready for validation, and activating it closes the parent quest
  • Governance body as a setup section: mandates with roles, seats, term length and rotation, and setup is not complete without it
  • Mandate votes: an owner, an admin or the acting chair opens a vote once setup is complete; the term starts at approval
  • Term end: holders are reminded before a mandate expires, with the succession vote prefilled and the rotation rule applied
  • Child workspaces inherit the parent's governance body, and the origin is shown
  • Invite from the network: a sub-workspace picks candidates from its ancestor workspaces, and a workspace can admit only members of its parent network
  • Shareable invite links with a role, an expiry and a quota, and a join page for them
  • Academy: a credential for every completed course, a quiz as completion requirement, evidence, certificate and badge on the course page, badges as profile achievements
  • A ProcesOS 3.0 course seeded from the rewritten platform and framework articles
  • Banking provider registry with Revolut Business: encrypted settings, a test endpoint and a signed webhook receiver
  • Operator ledger: transactions are ingested by webhook and catch-up, Stripe payouts are mirrored and reconciled, with an admin finance page
  • Payout accounts and payment orders with four-eyes approval, handed to the provider and completed through the ledger
  • Bundle products are unfolded into their parts at checkout and in the subscription webhook
  • LkSG rulebook with a framework entry and a supply-chain example template
  • Workspace type is chosen in the creation stepper

Changed

  • Quest projection reads the frozen AS-IS version the claim names rather than the working diagram
  • Publishing a quest sets it to published, so offers can be accepted right away
  • Status and tone colours come from declared theme tokens across the interface
  • Indexes on XP, skill progress and notifications keep those pages quick as they grow

Fixed

  • The audience hub shows its taglines
  • The sign-in panel scrolls on short viewports, and the quest bubbles on the login map keep clear of it
  • A capability in a terminal status carries no health warning, and gate redirects keep the request method
v3.3.0

Priced per Workspace

Per-workspace list prices, existing subscriptions keep their rate, a referral programme credits the people who bring a workspace in, and every landing page opens with the problem it solves.

Added

  • A referral programme: an initiator's code is entered at checkout, and 20 percent of the referred workspace's net plan price is credited to a workspace of the initiator for twelve months, never paid out and never combined with a founding place
  • A referral panel in the billing settings: the own code, where it pays into, the credits it has earned, and a code field on the upgrade with a plain explanation of who receives what
  • An operator ledger for referrals under Settings, with the rows that need a decision listed first and a one-way block for a code
  • An Advocate credential, minted once in the initiator's personal workspace on the first credit
  • A per-person view on the pricing page: what Team & Scale and the Pilot Set cost each month for 5, 20 and 50 people, divided from the list price
  • Founding places, ten per kind of workspace, with a live counter on the pricing page and an honest closed state
  • The Pilot Set: Professional plus four modules at 219 a month, next to the Full Platform at 290
  • The partner page names four kinds of partner, mappers, implementers, solution providers and initiators, with what each does, how they qualify and how they are paid, plus an illustrated flow with a video slot
  • The FAIR Passport explained on the academia page with a four-pillar switch, and a stewarded-ecosystems story told through one forester and one summer
  • The governance view rebuilt around a decision queue, coverage and attention, with AI systems staged by risk class and IT systems as first-class subjects
  • IT systems, data objects and resources as cockpit surfaces, with a signpost from the settings for personal workspaces
  • Personal data export and account retirement, and the RETIRE process of workspace_core that runs the deletion cascade
  • End-event completion credentials: a confirmed end event hands the lane's members a badge, confirmed only by the lane's responsible
  • Workspace OCR with your own key, next to the AI and voice keys
  • A process maturity fact on solutions, and a submission floor that keeps process-less listings readable as legacy IT systems

Changed

  • List prices raised to what the category charges. Subscriptions that already exist keep the rate they were signed at
  • AI Studio includes a fair-use quota of 500,000 tokens per workspace and month, with one notice at 80 percent. Workspaces with several members share the one budget, and a workspace answering with its own key is not metered
  • The Starter's three validated processes are a standing count: deleting or archiving a process frees its place again
  • Every amount the platform shows comes from one price list, and bundle savings are computed from it
  • Every audience landing page opens with the question it answers, links say where they lead, and no link drops a visitor into the editor or a sign-in unannounced
  • The ecosystem map and its detail panel share one stage, each actor's panel opens with a question, and a chosen actor is no longer remembered between visits
  • Landing copy across all pages rewritten to one thought per sentence, with jargon spelled out or dropped
  • The pricing page says how modules are bought: choose a workspace plan first, then add modules in that workspace's billing settings
  • The voice consent dialog names the chat model and its operator, not only the voice provider
  • Quest matching evaluates a candidate's language and location, and an empty profile field reads as not evaluable rather than unmet
  • A linked data object takes its name and fields from the library at render time, so a renamed object shows its current name in every diagram
  • Markdown renders through one canonical sanitizer everywhere it appears

Fixed

  • Team & Scale states the price of an additional workspace from the price list, so it cannot go stale in translation
  • Bundle cards show a saving only when the prices support it
  • Switching theme or language on a pricing anchor keeps the section in view
  • The marketing header stays on the Partners and Solutions pages
  • A draft or archived capability template is served only to its author and curators, on every sub-route
  • A provider's cost basis stays private to the provider, including for demo visitors of the cockpit
  • A pool without lanes reaches the governance tab as its own lane
  • A minted credential names the issuing workspace by its name
  • The Back button after a sign-out returns to a page rather than an error, and the error page offers a working sign-in link in every language
  • The editor shell fits the viewport on small screens, the toolbar folds, and dialogs stay reachable
  • Upload limits are announced at boot from the configured value
  • Blueprints open on the marketplace tab that exists

Removed

  • Governance Suite and Sustainability Pack, two narrow bundles whose discount had drifted to between 12 and 16 percent
  • The stand-alone blueprints page, which duplicated the marketplace
v3.2.0

Focus & Stewardship

Workspaces focus their own sectors' capabilities, stewards earn standing for curation work, relations get a hands-on editor, and linked datarooms become usable right from the forum.

Added

  • Workspace sector declaration with a focused capability tree: your sectors in the foreground, everything else dimmed but explorable
  • Domain segment health on the journey and workspace wheels — a named blocking capability with its standing duration, and an honest "nothing measured yet" state
  • Category stewards: the foundational and enabling catalogue layers can be assigned steward workspaces, with the same strict curation gate as sectors
  • Stewardship XP with a fixed published economy, and honorific per-scope stewardship credentials minted on first publish
  • A relation editor on the capability catalogue: typed add-relation dialog with evidence, drag-to-link on the tree, keyboard-accessible throughout
  • Forum-linked datarooms: upload from the thread, inline PDF viewing, and sanitized Markdown rendering under the dataroom's access policy
  • A licensed-instance workspace-creation flow: license-provided features instead of prices, and basics inherited from an admin-designated primary workspace
  • The business backoffice capability pack: ten seedable templates from bookkeeping to contract management, with ESCO-verified skills and a month-end-close blueprint

Changed

  • Withdrawing a capability relation deprecates it with full provenance — relations are never deleted
  • Cycle refusals in the relation editor quote the exact loop so the fix is obvious
  • Passport review routing now honors category stewards automatically wherever sector stewards already applied

Fixed

  • Workspace connection requests reach the receiving side: admins are notified, the incoming request shows accept and decline where the receiving workspace looks, and a pending-requests banner appears on the cockpit
  • Connection notifications arrive in the reader's language instead of hardcoded English
  • The sector taxonomy admin panel and curation surfaces resolve stewards for every scope through one shared index
v3.1.0

Capability Graph & Sector Stewardship

The capability model becomes a curated graph: typed relations, sector-steward reviews for FAIR Passports, and personal career paths derived from ESCO occupations.

Added

  • Curated capability relations: all thirteen relation kinds as typed edges with ranks, qualifiers and references, editable through one guarded write path
  • JSON-LD publication of the capability catalogue under a versioned UCX namespace, with content negotiation on the catalogue, template and sector APIs
  • Personal starter catalogue: adoptable capabilities across all seven life domains, seeded per instance by an operator command
  • Sector fan-out on the personal journey: professions derived from ESCO occupations, adopt-to-curate shared catalogue entries, and a seven-rung responsibility ladder with nine worked professions
  • FAIR Passport sector reviews: steward queues, a changes-requested round trip with mandatory written reasons, resubmission from the solution page, and notifications on both sides
  • Quality, capacity and utilization conditions on capability instances, with capacity recorded as declared availability — never compared or scored
  • Domain wheels: one curated grouping axis per capability target type, from persons to nations, with sector-projected sub-segments
  • Five-value statement verdicts on personal capability requirements, so nearly-met setups read as what they are
  • A sixth global handbook article explaining who tends the capability catalogue

Changed

  • Exit and repatriation are sourcing options with their own panel on the template detail — no longer graph relations
  • Seeded catalogue entries carry a steward who can curate them; sector taxonomy upgraded to ISIC Rev. 5 with typed crosswalk predicates
  • The elemental grouping axis left the public capability model; the five roots continue as the community capability slice under forward-compatible aliases
  • Seed skill requirements resolve against the pinned ESCO release before anything is written — an unresolvable skill stops the seeder
  • The governance tab counts every stakeholder-validated maturity level from three upward and renders one honest legend
  • Login and network maps draw a self-contained stage, preparing fully self-hosted basemap tiles

Fixed

  • The workspace-creation wizard advances to the next step immediately after each answer is saved
  • Capability validation refusals surface their reason in the cockpit instead of failing silently, and the gtt:view / gtt:validate permissions are enforced on every route that declares them
  • An expired FAIR Passport also retires the solution's certification mark
  • A personal workspace can no longer be made the parent of another workspace
  • The cockpit tree tab reads its own workspace's tree with bounded queries
  • Workspace-scoped API keys act only within the workspace they are bound to when confirming taxonomy crosswalks
v3.0.1

Reachable Again

The app bar follows you onto workspace profiles instead of stranding you there, and you can finally create the personal dataroom the CV import always expected to find.

Added

  • Datarooms in personal settings: create and configure the encrypted rooms your evidence import has always expected, on any plan

Fixed

  • The app bar stays with you on public workspace profiles, so opening a workspace from the directory no longer strands you there
  • The app bar no longer overflows on medium windows, where its links were allowed to shrink out of the row instead of hiding
v3.0.0

The Personal Side

Every person gets a real place of their own: a capability journey along seven life domains, a first quest that ends in a genuinely activated capability, and solutions that travel on a FAIR passport.

Added

  • Your journey: a personal capability graph along seven curated life domains, with domain progress read from your real activity
  • The first quest: four honest steps from signup to your first activated capability, worth 425 XP across the arc — each milestone paid once, forever
  • Academy lesson player with per-lesson progress, and an eight-lesson orientation course in English and German, editable per instance
  • All ten capability activation requirements are now evaluable — experience, location, working languages, delivery mode, capacity and an NCTB-aligned identity floor included
  • Personal equipment: record your systems, machines and frameworks and bind them to your capabilities as honest self-declarations
  • Capability milestone XP: adopting pays 25, activating pays 150, through the same audited economy as everything else
  • Solutions declare their audience and their capability output — who they are for, and what they provide, improve, require, use or activate
  • FAIR passport machinery: tiers, applications, appeals with reminders, examiners, readiness, expiry and countersignature integrity
  • Workspace hierarchy: consent-guarded parent and child links, capabilities that spin out into sub-workspaces, bottom-up capability and cost rollups
  • AI stack: OmniRoute completion provider, document OCR, a speech provider per capability, and a provenance line naming who runs the model

Changed

  • FAIR is the passport status, not the entity name — the registry lists solutions, and the badge renders exactly when a passport is issued
  • The personal area is findable: diagrams live in the personal shell, the dashboard links your journey, and the first quest sits where a new person lands
  • Onboarding settles into three skippable steps with a handover screen, profile XP paid on facts, and a light-or-dark choice before the first paint
  • Four themes become two maintained themes plus the whitelabel CI, whose brand colour is derived to a readable contrast
  • Academy certifications are the evidence gate for skills: completing a course now issues the certificate a crossed skill-tree threshold promised

Fixed

  • Courses can actually be completed: the lesson player surfaces the completion path that existed server-side with no way to reach it
  • 96 undeclared design tokens now resolve instead of rendering hard text-coloured borders, and a new test keeps the class of bug visible
  • The status palette uses one declared name per colour instead of five undeclared spellings
  • Capability nodes export correctly to SVG and PNG instead of leaving holes
  • Artifact updates are scoped to their workspace exactly like every sibling write
v2.20.0

Guided Setup, Incentives & Claims

Creating a workspace is now a linkable, resumable six-step flow. The XP economy moves into a cockpit Incentives area, and people can submit claims for verification across workspaces.

Added

  • Workspace creation runs as six linkable steps with its own routes, so a setup can be shared, left and resumed
  • Unfinished workspace setups are listed for 30 days with the step each stopped at, and can be picked up where they ended
  • Provisioning shows the phases it has completed and reports what the template actually created once it is done
  • Ecosystem roles are grouped by archetype during setup, and every template states how many documents it brings
  • Incentives is a cockpit area with five tabs — Rules, Achievements, Rewards, Impact and Skills
  • The rules tab makes the XP economy readable: what a bounty, a maturity level and a credential type are each worth
  • People can submit a confirmed claim to another workspace, carrying labels and types only
  • Workspace members with verification rights can decide a submitted claim, recorded as a separate result

Changed

  • Incentives moved out of the workspace settings and into the cockpit, next to the capability tree and the network graph
  • Skill statistics count people rather than progress records, so member totals, averages and top lists match the workspace roster
  • Averages are shown only where the group is large enough to support them, and always name the population they came from
  • Personal offerings can only be activated once their setup is validated, and are presented in three groups

Fixed

  • Every message catalogue that ships with a release is compiled into it, so newly added areas render their text on first load
  • A person working across several departments counts once towards a skill, with the level their combined experience earns
v2.19.1

Changelog Clarity

The release history now reads as release notes rather than as internal bug-tracker entries, with every date and technical detail left intact.

Changed

  • The fixed entries of releases v1.0.1 through v2.9.0 describe the behaviour that now holds, with their technical substance, magnitudes and publication dates unchanged
  • Three changelog entries with no observable effect on the product were removed
v2.19.0

Identity Assurance & Credential Verification

Identity assurance becomes its own axis I0-I4, credential claims can now be verified, and a failing index migration no longer blocks a deploy.

Added

  • Identity assurance ladder I0-I4 as its own axis, derived from append-only verification events so a revocation actually lowers it
  • Credential verification runs that record what was checked, and name the seven checks this instance cannot yet perform
  • A "what we did not check" block on verified credentials, so an absent check can never be mistaken for a passed one
  • The seven personal claim kinds now render as translated labels instead of raw slugs

Changed

  • Identity assurance is global to the person rather than workspace-scoped, so eligibility can be judged across workspaces
  • Migrations create indexes through a shared helper that matches on the key pattern and reports differing options instead of replacing them

Fixed

  • Schema migrations recognise an equivalent index that carries a different name, so an upgrade applies cleanly on instances whose collections were seeded by earlier tooling
v2.18.0

The FAIR Passport, and Capabilities That Know Who They Need

A complete FAIR Passport pipeline — four pillars, examiners, appeals, issuance — and capability nodes that can finally name the people, skills and occupations they require.

Added

  • FAIR Passport pipeline: four pillars, rubric as configuration, three scoring runs with the median per pillar, and run provenance
  • Examiner path: allocation with a conflict-of-interest block, counter-signing that holds up issuance, appeals with a named decision-maker and a deadline, and the FFDE role that edits evidence but never scores
  • Passport issuance with two-year validity, an expiry sweep and a public register; eligibility gate on TRL and four validated process views
  • Workspace Compass — questions, ranked results with their reasons, and a join request
  • Skill-tree builder with draft trees, node rewards, reward types and achievements
  • Workforce roles on a capability definition, mapped to ESCO occupations, with per-target overrides
  • Skill requirements suggested from a node's occupations — provenance-tagged, curator-decided, never applied automatically
  • Node card sections "Skills and credentials" and "Roles and authority", with skill names instead of identifiers in en/de/es
  • Person ↔ capability matcher with six named evidence states and no score
  • Workforce-transition map per capability, attachable as evidence for its workforce assessment
  • Capability-derived learning pathway: course draft, Setup and Run participation as components, Open Badges 3.0 export with ESCO alignment
  • ESCO occupation registry (~3 000 occupations with ISCO-08 groups and their essential and optional skills) and a runtime load path for the ESCO dataset
  • Assurance ladder A0–A4 on claims and evidence generally; ISIC / COFOG / SDG crosswalks on the sector taxonomy; distributional social-outcome fields on human-outcome statements

Changed

  • No composite score is shown without its breakdown — the FAIR transparency rule, applied to readiness as well
  • Readiness is a statement of named met / unmet / cannot-be-evaluated lists with blocking flags; the percentage is removed
  • Workspace skill checks honour each person's release level for their performance data; withheld data is reported as a count, never as a zero
  • Human capability requirements can name the skills they ask for, and every skill row on a capability records where it came from
  • The typed edge vocabulary between capabilities now matches the eleven Geneva relation types
  • The discussion list reads as a table, and a thread's pipeline moved into a header band
  • The workspace overview is three columns, with maturity shown next to the lever that moves it

Fixed

  • Skill requirements on a capability were permanently unsatisfiable — the check read fields nothing ever wrote and reported the result as "not earned yet"
  • Credential requirements on a capability never matched, because the check looked for the slug in a place credentials do not carry it
  • A capability that could not be evaluated no longer counts as one that failed
  • Achievements are granted again — the evaluator, the id space and a missing branch

Removed

  • The numeric readiness score and its breakdown
v2.17.0

Finding and Joining a Workspace

Workspaces can now be discovered, asked to join, and approved from the members list — and a new personal onboarding screen shows a first-time visitor where to start.

Added

  • Workspaces can appear in a public `/explore` directory and accept join requests, both opt-in per workspace
  • Join requests notify the people who can decide, and can be approved from the bell or from an approval queue in the members list
  • A personal onboarding screen for people who have just signed up
  • The capability detail panel shows required, optional, alternative and supporting dependencies, with blocking kinds kept apart from the rest
  • "Where do you want to go?" — pick a capability and see the missing prerequisites as an ordered route
  • A forum thread row shows whether the discussion was analysed, drafted or proposed, and marks the ones that have stalled
  • Setup, Run, Change and Retire processes are part of a capability's data model
  • The FAIR passport rubric is configuration rather than code

Changed

  • ProcesOS is now proprietary software; tech reform GmbH holds all rights
  • The network map follows the active theme and explains its pins, with the location form beside the map instead of below it
  • A manual FAIR score override can be removed, and the computed score stays visible underneath it
  • Both forum filter rows name the axis they filter, and every label goes through translation

Fixed

  • The thread analysis renders emphasis instead of printing the asterisks around it
  • `--bg-active` and its foreground are declared in every theme — the "you are here" background rendered nothing at all
  • Settings navigation groups are announced with the entries they head, and the current entry is marked as current
  • Five status colours that failed contrast on the light theme now use tones that hold on every theme
  • An industry no longer matches a sector because its name contains the word
v2.16.0

Capability Authoring & Access Control

Capabilities can now be created and adapted from the catalogue itself, and three surfaces that answered to a login alone now ask who is actually allowed.

Added

  • Create and adapt capability definitions directly from the catalogue and from any capability page, with category as a required choice
  • Filter the capability catalogue by category, and see it on each card
  • Curating the role-profile catalogue is now a separate, delegable permission rather than part of instance administration
  • A lane map beside the capability force graph, with the selection surviving the switch between the two
  • Workspace settings are grouped into five sections instead of one long list

Changed

  • Adapting a capability now links the processes that already referenced it, instead of waiting for each diagram's next save
  • Moving a diagram to another workspace takes its process-role links along, rather than leaving them behind in the old one
  • Saving in workspace settings names the section that was written, so two saves are no longer indistinguishable

Fixed

  • Certification submissions, estimates and partner requests could be read or raised for diagrams outside the caller's workspaces
  • Anyone signed in could add an entry to the global role-profile catalogue and publish it
  • The capability picker when publishing a marketplace listing was always empty
  • A capability could be created with an address that nothing could open afterwards
  • Listings could be published at a negative price
v2.15.0

Capabilities Meet Their Processes

A capability node now points at the diagrams that build and run it, validating one activates it, and the cost it carries is finally on screen instead of behind an endpoint nobody called.

Added

  • A capability's `uses` relation binds to the processes that build and run it, with the bound diagram openable from the properties panel and unbacked claims flagged during validation
  • Validating a capability's setup diagram activates the capability — a direct link no longer needs a blueprint
  • The capability cost basis is visible on the node, in its detail view and in the workspace cockpit, with an explicit recompute
  • Saving a diagram records which capabilities it touches, so the link between a process and a capability no longer has to be declared by hand
  • Walkthrough steps for capability nodes carry description, status and the systems behind them
  • A collapsed subprocess or call activity can reference another workspace diagram and open it
  • Confirming a claim now writes the assertion it derives

Changed

  • Writing to the global capability catalogue now requires the `gttTemplates:manage` permission
  • The five capability element colours come from one table, with a separate fill tone for surfaces carrying white text — measured against every theme surface rather than chosen by eye
  • The home page hero image loads eagerly; it is the element the page is judged on

Fixed

  • The readiness list names which dependency is missing instead of repeating "Missing dependency"
  • Capability nodes no longer sever the sequence graph in the walkthrough
  • Numeric entities written by the XML serializer are decoded in business-metric attributes, so a value round-trips as itself
v2.14.0

Personal Import & Cost Transparency

Bring a CV or Europass file into your own workspace on the free tier, trace what a process costs task by task, and reach every cockpit tab before a single process is validated.

Added

  • Personal Import page for Europass XML/JSON, LinkedIn CSV exports and plain-text CVs, with per-claim confirm, correct or discard
  • Internal datarooms on the free tier with a 500 MB quota, and pasted text stored as Markdown
  • Cost Breakdown tab in the analytics dialog showing the calculation task by task, with branch and execution weight
  • Editor command that opens the analytics dialog directly on the cost breakdown, reachable from the command palette
  • Workspace quest rows now show deadline, evidence type, scope and XP instead of only a slug and a price

Changed

  • All cost figures come from a single engine; the parallel implementation that never reached the editor's data model is gone
  • The cost view names its own blind spots — unresolved lane rates and truncated rework loops are stated rather than costed as zero

Fixed

  • A logged-in user with no membership could name another workspace and have its stored AI credentials used on their behalf
  • The cockpit no longer hides its tab bar and all ten tabs when a workspace has no validated process
  • Removing a treasury address is no longer blocked by unpaid internal work orders, which pledge nothing and never close
  • The ecosystem map is reachable again, and its anchor links work
  • Import and review routes refuse API-key authentication, which cannot be scoped to a single workspace there
  • Diagram SVG and PNG export now renders in the browser and reports a failure instead of quietly producing nothing
  • Long AI assistant conversations scroll inside their panel rather than pushing the editor out of shape
v2.13.0

Federated Events & Change Execution

Instances now exchange signed, ordered events. Changes can be forked, approved and promoted over HTTP. And the landing page finally shows the product.

Added

  • Federation exchanges signed events between instances: an append-only store with idempotency and ordering, an authenticated inbound route, and an outbound queue with its drainer
  • Revocation propagates across federated instances — by construction it can withdraw a grant but never issue one
  • The federation event queue is readable by an operator, so a stuck exchange can be diagnosed instead of guessed at
  • UCX Change Execution over HTTP: fork a process, approve it, promote it
  • Personal and household workspaces carry their own capability graphs
  • ⌘K opens a command palette in the editor, filled with the editor's own verbs
  • Capability nodes participate in sequence flows, and the AI wizard can create and read them
  • Workspaces see their own quest drafts in the management tab
  • An admin panel names the registry entry the landing page exhibits — previously only reachable by editing the database
  • The landing page shows real product screenshots in all five frames: editor with the assistant, cost and path analysis, the exchange, and the workspace cockpit

Changed

  • Each quest row shows one readable status, derived from what the quest actually declares
  • Quest generation is locked across replicas, and generated rows say so

Fixed

  • The marketplace refuses self-offers on quests on the server, not only in the interface
  • A household with no members is reported as a defect instead of rendering as an empty one
  • An absent capability list is no longer read as an empty list
v2.12.0

Workspace Cockpit & Capability Flows

The workspace cockpit gains sovereignty and FAIR views, capability nodes join sequence flows in the editor, and the Academy opens its learner and author views.

Added

  • Workspace cockpit: sovereignty view across four jurisdictions, a FAIR matrix per system, and FAIR scale / jurisdiction buckets as token-bound bands
  • Capability nodes participate in sequence flows — splice them into a flow or append them from any task, event or gateway
  • Editor navigation in three levels (mode · view · action) with a process state strip in the toolbar
  • Academy: personal "for you" view with level gain and learning progress, plus an author view exposing the publish checklist
  • Jurisdiction stamps record where a process is compliant as-is, backed by a declarative compliance rulebook format
  • UCX lifecycle: handover packages, reopen triggers, and capability evidence rolled up from diagrams
  • Federation administration is separately delegable via new federation permissions
  • Wizzy, the assistant, has a real drawn figure

Changed

  • The theme picker offers the two maintained themes, and the theme cookie is validated instead of trusted
  • All themes gained the semantic tokens components were faking; marketplace surfaces no longer assume a dark theme
  • Every number on the marketplace states what it counts, using one shared label module
  • Stranded course enrollments are marked instead of deleted

Fixed

  • The employer's view of a member's level is workspace-isolated again
  • Ecosystem map edges meet the 3:1 contrast requirement where they are drawn
  • Admins can set which registry entry the landing page exhibits — the setting existed but had no UI
v2.11.0

Federation & a Rebuilt Front Door

Instances can now agree to work together under signed, scoped terms. And the public site was rebuilt from the ground up — nine chapters, two new explainer pages, one honest companion.

Added

  • Federation agreements between instances: mutual terms, key challenge, signed handshake, and an admin panel over the whole state
  • Inbound and outbound federated calls are gated on an active agreement, persisted and audited
  • `/capabilities` — a public explainer for the Universal Capability Exchange, with a shared status vocabulary bound to the product's own node states
  • `/ecosystem` — eleven actors on a map with their connections and a second view of who is accountable to whom
  • The registry chapter reads a real published solution: setup and run as a pair, six SC4 dimensions, verification level, and where each figure comes from
  • Orbit replaces the landing chat — read-only enforced server-side, with a ring that reports the knowledge actually used
  • Capability versioning: fork, approve and promote without destroying the previous state
  • Workspace sector curation, an incoming-steward worklist, and per-instance stewardship
  • A first-class workspace type axis, indexed, driving trial gates and licence lists
  • Academy articles carry translated prose through a locale sidecar

Changed

  • The landing page is driven by one ordered chapter array — adding a chapter is a single line, and numbering follows automatically
  • Typography, spacing, radii and motion are theme tokens; fonts are self-hosted with no third-party CDN request
  • The workspace overview was rebuilt as headline, metric strip and open points
  • The process map is now the default view of the diagram listing
  • Notification bodies render from typed payloads in the reader's own language
  • Wizzy's readiness check names consequences instead of deficiencies and no longer blocks the start
  • Wizzy's confirmation dialog shows a diff — target, before, after and consequence — and offers a per-turn undo
  • The wand signals a real mutation only; reads, searches and simulations never trigger it
  • Paraglide compiles per locale and the build is gated on a memory budget

Fixed

  • A logged-in user could read and write another workspace's resource library through a caller-supplied workspace id
  • SC4 environmental sustainability could never reach 5.0 and scalability silently saturated — every dimension now sums to the full scale, and scores from different weightings are never subtracted from one another
  • Unpublished academy articles were reachable, and workspace handbooks could enter the global knowledge index
  • Level rewards were counted across workspaces instead of the caller's own
  • Course completion XP was credited outside the real ledger
  • URL-localized routes were linked without their locale prefix, sending German visitors to the English page
  • Light theme muted text was green-tinted and below the WCAG AA threshold
  • Several cockpit counters were structurally wrong, and failing card queries took the whole page with them
  • A forked capability template did not resolve on the read path
  • The shipped env bundle was not covered by the ignore rules
v2.10.0

The Capability Catalogue

The Universal Capability Exchange arrives: a forkable catalogue of capability definitions, assessments across thirteen dimensions that never collapse into one score, and licence-wide module control.

Added

  • Universal Capability Exchange: a reference catalogue of capability definitions that nations, sectors and institutions can fork across five adaptation layers without mutating the common node
  • Capability type vocabulary — a stable classification of what a capability is, held separate from the node that defines it, so a national fork never breaks the shared vocabulary
  • Six mandatory human fields on every capability definition, including the judgement that must remain human and the outcomes that are never acceptable, carried as data in any language
  • Capability assessments across thirteen separate dimensions with supporting evidence, and eight red flags that route a capability to review or refusal regardless of how strong the rest of the picture looks
  • Fork write path: create a fork at a lower adaptation layer and override local metadata, with lineage fields refused rather than silently ignored
  • Module switches under an enterprise licence: disable a licensed module instance-wide or for a single workspace, without touching what was bought through Stripe

Changed

  • Marketplace acceptance now enforces credential and skill requirements instead of merely recording them
  • The Cardano escrow validator ships compiled with a script-hash drift tripwire, and mainnet operation is refused without an explicitly pinned escrow address
  • Testnet and mainnet signing keys are separated structurally, so a misconfiguration cannot quietly reach the wrong network

Fixed

  • Reference catalogue entries no longer count towards a workspace's capability maturity, dashboards or unlock cascades — publishing a catalogue no longer lowers everyone's score
  • Creating a template through the raw write path can no longer bypass the fork rules
  • The forum now keeps the active workspace in step with the thread being viewed
  • Under an enterprise licence the admin table showed workspaces as having no modules when they in fact had all of them; it now reports the effective set and where it comes from
v2.9.0

Shell, People and a Guided Start

A global bar above every view with notifications grouped by urgency, the Departments tab becomes a real People directory, and new workspaces get a guided setup instead of a modal.

Added

  • A global bar above every app view: my area, workspaces, marketplace, academy and forum, plus a workspace switcher, ⌘K command palette, level and notification bell
  • Notifications grouped by urgency with counted filters, each line naming the workspace it came from
  • People directory in the workspace cockpit: lanes resolved to people, contacts, process involvement and key-person risk
  • Three-step release levels for performance data — private, workspace or network — controlled by each person for themselves
  • Guided workspace setup: a resumable flow with credentials, invitations, process-map stubs and data-object capture
  • Instance-wide artifact registry with cost attribution per cost object and per licence line
  • Bulk invite and update of members from a spreadsheet
  • Marketplace browse-and-inspect layout with facet counts and a way out of an empty result

Changed

  • The personal dashboard is one screen without tabs; what waits for you is sorted by deadline across types, and workspaces are ordered by open items rather than alphabetically
  • A person's level is a number. Tier names no longer appear next to it, though the tier colour stays
  • Course XP is derived from the incentive table instead of being entered freely
  • Skill cards became rows, and the two values nobody could interpret are gone

Fixed

  • Draft courses are readable by slug, enrollable and publishable only once they carry at least one lesson
  • Custom credential gates can be satisfied — the credential slug is persisted with the gate
  • The Spanish translation gap is closed, with a guard against it reopening
  • Trial expiry is written compare-and-swap, so concurrent requests cannot clobber it
  • The vote board renders proposals that carry an ROI date
  • One live edge per workspace pair in the workspace graph, instead of one per workspace
  • Dataroom feature gates resolve through the licensed subscription
  • A failing migration is isolated so the rest still apply, and a duplicate migration id is refused rather than silently skipping one
  • Every sample workspace exports, and template references resolve
v2.8.0

A Seeded Ecosystem & Cross-Workspace Seals

Eight sample workspaces to adopt as templates, processes that carry a seal across workspace borders, and capabilities that now activate on Setup — where they are actually created.

Added

  • Eight connected sample workspaces seed as a working ecosystem — bakery, mobility association, self-sustaining village, stewarded ecosystem, solution provider, transformation consultancy, municipality and issuing authority — with cross-workspace adoption, a quest and a collaboration link between them
  • Four of the samples are adoptable workspace templates, so a new workspace can start from a solution provider or a municipality instead of an empty page
  • Cross-workspace process certification: request a seal from an issuing authority, which reviews the diagram in a shared collaboration space and certifies it against a named standard
  • Lane-staffing requests — a lane with a credential requirement must be staffed before a process can be certified, and an open request blocks the seal
  • Certification is a portable credential that travels with the process when its template is adopted
  • Lane roles carry their required credentials into exported templates, so an adopted process keeps its qualification bar
  • Administrators can seed the sample workspaces from the admin panel, including a reset that refreshes a demo instance
  • Soziale Fahrten licences can be bought through ProcesOS

Changed

  • Capabilities now activate when the Setup process is validated, not the Run process — a Run process cannot be validated for a capability that does not exist yet. Existing blueprints are migrated automatically, and anything ambiguous is left untouched and logged rather than guessed
  • Governance leads are seated in the shared collaboration workspace, so a head of compliance can actually review what their workspace was invited to
  • Exported templates keep the own pool and the partner pool distinguishable, so a two-party process survives adoption as a two-party process
  • Four eyes are enforced across workspace boundaries: whoever requests a certification cannot grant it

Fixed

  • Workspaces whose diagrams reference CMDB artifacts can be published as templates — the embedded database ids are rewritten into placeholders that resolve inside the manifest
  • A reference pointing at a record that was never exported is refused at publish time instead of shipping as a link to nothing
  • Every sample workspace exports cleanly through the anonymizer; random identifiers that leaked into exports were replaced with stable ones
  • VAT is requested during workspace checkout
  • The setup check degrades gracefully when the database is unreachable, instead of turning every route into a server error
v2.7.0

Languages Per Deployment & Voice Transparency

A deployment can now ship its own languages via LOCALES — declare generously, translate over time. And Wizzy shows which voice provider handles your audio before you opt in.

Added

  • Languages are configurable per deployment through a LOCALES environment variable — locales.config.mjs is the single source of truth and drives localized routes, the inlang project, and every locale list in the app
  • LOCALES travels the build path (GitHub Actions to Docker build argument), so an image carries the languages it was built for instead of falling back to the default set
  • Declared versus translated locales: a declared language reserves compiled capacity and gets AI narration and interview questions immediately, while the interface offers it only once its translations exist
  • Voice provider details before consent — provider name, who operates it, and how audio and text flow (speech-to-text discards audio, text-to-speech caches for 30 days), shown in the voice-interview consent dialog and behind an info trigger next to the voice controls
  • Voice configuration reports its scope, so you can see whether voice runs on your workspace's own provider or the instance-wide configuration

Changed

  • Accept-Language is parsed properly with quality values instead of matched against a whitelist
  • Language names appear as endonyms via the platform's display names (Kiswahili, Gikuyu, Dholuo); flags are reserved for the shipped languages, since a language is not a country
  • The language switcher, locale negotiation and the sitemap derive from the translated locale set; the AI narration and elicitation endpoints deliberately use the declared set

Fixed

  • Email templates fall back to the base locale for a language without its own templates, instead of failing outright
  • The sitemap emits URLs and hreflang alternates only for translated languages, so search engines are never pointed at English content under a foreign language code
v2.6.0

Whitelabel, Demo Mode & Verified Mandates

Instances can carry their own branding, open a public demo, and issue term-bounded committee mandates whose on-chain proof is actually verified — plus a broad security pass.

Added

  • Instance whitelabeling: name, logos, CI colors and a slim branded landing page, gated on the instance license
  • Demo mode: magic-link sign-in, read-only browsing of ecosystem workspaces, consent-based newsletter opt-in
  • Governance-role credentials for committee mandates (chair, vice-chair, delegate, member, observer) with terms, appointing body and declared interests
  • Data-driven workspace structure for governance bodies, including sub-committees, cross-cutting bodies, an organisation workspace and one workspace per summit
  • On-chain verification of minted credentials against the chain, with the network always labelled and an outage never devaluing a genuine proof
  • Versioned sector taxonomy with crosswalks, and sector tagging for diagrams, blueprints, artifacts and capability nodes
  • API keys can be issued with a read-only scope for partner integrations
  • Datarooms can be gated on committee membership, role and appointing body

Changed

  • API keys are now strictly limited to the workspace they were issued for; key authentication is refused on admin and key-management endpoints
  • Credentials whose validity or mandate term has passed are now reliably treated as expired
  • Pricing pages show curated, localized plan features again instead of raw internal feature names
  • The release banner no longer mixes languages, and the mobile navigation header is decluttered

Fixed

  • AI-generated report HTML is sanitized before it is rendered, printed or copied
  • The session cookie keeps its httpOnly protection — the session object is no longer serialized into the page payload
  • Workspace member and dataroom routes reject non-string filter input (query injection)
  • Credential issuing via the Andamio import requires a write permission, and the mint endpoint enforces membership
  • Module gates added to compliance, CMDB and marketplace mutations that were only gated in the client
  • Rate limit and token bound on the AI completion endpoint, hard timeout on the thumbnail worker, URL validation for avatars and the Odoo endpoint
  • A repair migration normalizes invalid capitalized element names in stored diagram XML, restoring BPMN 2.0 schema validity for exports
v2.5.0

Own Voice, Own Storage

Wizzy's voice can run fully self-hosted or on your workspace's own provider, datarooms gain decentralized Storj storage with uploads, and the workspace handbook reads your Odoo knowledge base.

Added

  • Self-hosted voice provider (Speaches) with locale-dependent TTS voices — German, English and Spanish each speak with a fitting voice, fully on-instance
  • Workspaces can bring their own voice provider (AI Studio module): keys encrypted in the workspace vault, per-workspace audio caches
  • Admin voice settings list available models and voices live from the configured provider
  • Storj as decentralized dataroom storage: credential-gated listing, shared-link downloads, and direct uploads from ProcesOS into the bound folder
  • Workspace handbook: reading surface and dashboard widget combining workspace articles with a bound Odoo knowledge subtree (read-only proxy)

Changed

  • Wizzy's spoken output follows the app language — each locale gets its own cached audio
  • Dataroom error handling distinguishes 'no access' (request access) from 'workspace connection broken' (admin issue) consistently across providers
v2.4.0

Sovereign Licensing & Voice Interviews

Dedicated instances can run on a signed offline license instead of per-workspace billing, and Wizzy now conducts the process interview by voice — with consented transcripts to the forum.

Added

  • Instance license mode: signed offline license key replaces Stripe entitlements instance-wide, with an enterprise-mode switch and admin license panel
  • Voice interview: Wizzy speaks elicitation questions, accepts push-to-talk answers and spoken confirmations, and posts a consented Q&A transcript to the linked forum thread
  • Capability-node listings bundle their Setup+Run processes; CMDB artifact listings carry an ISMS evaluation profile and link to the capability nodes they enable
  • Capability nodes list the marketplace solutions that satisfy them, and adopted solutions arrive in your CMDB with their published ratings
  • Acquisition cost basis for capability nodes derived from Setup/Run processes, with normalized currency and rate units
  • Run-evidence KPI triggers: measured outputs outside thresholds reopen the setup loop
  • Tasks can reference existing runbooks and automations directly from the BPMN properties panel

Changed

  • Marketplace categories renamed to the operating-layer vocabulary: Capability Node, CMDB Artifact, Workspace Blueprint; standalone process blueprints are retired from browsing
  • Blueprints gained a full lifecycle: workspace scoping, draft state, archiving — nothing is published forever by accident anymore
  • Business-metric fields are normalized and validated on diagram save (currencies, duration units, structured JSON fields)

Fixed

  • Workspace identity creation works on a fresh deployment — the signing library's ESM build is restored at install time
  • The AI diagram assistant reports what went wrong on large batch builds: invalid element types are rejected with clear feedback, broken connections abort early, and the model is told the reason
  • Adopting a capability creates its dependent nodes
  • A diagram cloned from an adapted capability carries its content
  • The changelog shows the running app version, and the voice settings panel opens in the admin dashboard
v2.3.0

Wizzy Speaks, Free Drive Datarooms & Licence Compliance

Wizzy gets a voice — provider-agnostic speech in the editor and dashboard. Google Drive datarooms are now free for every workspace, plus a licence-compliance pass.

Added

  • Wizzy voice layer — have answers read aloud and speak instead of typing, in both the editor assistant and the dashboard guide
  • Provider-agnostic voice configuration decoupled from the chat provider — Mistral (Voxtral) and any OpenAI-compatible endpoint, including self-hosted inference servers; configured under AI Settings and applied without a restart
  • Push-to-talk with hold-to-talk or click-to-toggle, capped at 90 seconds; the transcript lands editable in the composer and is never sent automatically
  • Voice usage metering per user and per workspace (characters spoken, seconds transcribed), with cached playback excluded from billing
  • OSS licence inventory plus a generator, so the dependency licence position can be audited rather than assumed

Changed

  • Google Drive datarooms are no longer a paid module — free tier at zero cost, toggleable per workspace by an admin
  • AGPL-licensed dependencies replaced, the bpmn.io watermark restored as the bpmn-js licence requires, and a NOTICE file now ships with the application
  • Spoken audio is never stored — only the resulting transcript is; synthesised speech is cached so repeat playback costs nothing

Fixed

  • Walkthrough interviews: answers write through, a declined or negative answer is remembered instead of being asked again, and questions respect the semantics of the task they belong to
  • Drive datarooms: folders are reachable, credential pickers show the workspace's real credentials, and forum links are scoped to the folder
  • The walkthrough's Back button is enabled whenever there is a step to return to
v2.2.0

Guided Walkthrough Mode, Google Drive Datarooms & Sovereign Hosting

Turn any diagram into a narrated walkthrough that also captures stakeholder data, back credential-gated datarooms with Google Drive, plus a landing overhaul and the move to sovereign self-hosting.

Added

  • Guided Walkthrough — Presentation mode: a read-only token stepper that walks a diagram element by element with one-sentence Wizzy narration (generated once per version, cached) and a plain-language data panel; you advance with Next and choose branches at gateways
  • Guided Walkthrough — Elicitation mode: the stepper becomes an interview with a deterministic gap engine and Wizzy Q&A; answers are captured as provenance-tagged, staged claims, never written silently
  • Guided Walkthrough — CMDB auto-capture: systems named in passing are fuzzy-matched against existing artifacts (match offers a binding, miss offers a draft); drafts are excluded from compliance/scope analysis until a steward confirms, merges, or deletes them
  • Google Drive as an external dataroom storage provider — connect a workspace Google account, back a credential-gated dataroom with a Drive folder, list and preview files under the same access policy (including lane-based gating)
  • Forum threads can link datarooms, carrying a credential-gated document space alongside the discussion
  • Featured login providers with a "more options" dropdown on the sign-in screen
  • URL-localized marketing routes (/de, /es) with hreflang

Changed

  • Landing page restructured around the Process-OS narrative with one canonical value proposition across en/de/es, plus brand-story videos on the business and government pages
  • ProcesOS now runs on sovereign, self-hosted EU infrastructure (website, application, and the Veridian/KERI identity agent) instead of a managed hosting provider

Fixed

  • Andamio integration validated and hardened
  • Comment requests raised from an element click carry the workspace id, so they resolve in the right workspace
  • An unverified email address is redirected to verification
v2.1.0

Wizzy on the Dashboard, hiop.io DataOps & a Dockable Assistant

Wizzy comes to the dashboard, a new hiop.io DataOps module ingests real process metrics, self-hosted OpenAI-compatible AI endpoints are supported, plus a dockable assistant and a security pass.

Added

  • Wizzy on the personal dashboard — a user-centric AI guide that reads your workspace context, role, open tasks, maturity standing and recommended next capabilities, with read-only tools scoped to your workspaces and quick-actions (What's on my plate / What to map next / How are my workspaces doing)
  • hiop.io DataOps workspace module (ingest MVP) — push real cost / duration / frequency / volume metrics to a workspace-scoped, API-key-authenticated ingest endpoint; measured values surface next to modeled estimates in process cost analysis
  • Mandatory DataObject classification on every hiop ingest binding — each transfer is GDPR-classified via the existing DataObject library; special-category bindings blocked, personal-data bindings flagged
  • Idempotent metric delivery (by delivery ID) plus an integration panel that generates a copy-paste hiop.io pipeline snippet
  • Resizable and dockable Wizzy chat panel in the editor
  • Support for self-hosted, OpenAI-compatible AI endpoints via a new AI_BASE_URL setting — run against your own model server instead of a hosted provider

Changed

  • The editor's Wizzy chat now bounds its height so long conversations scroll inside the panel instead of expanding the layout

Fixed

  • AI-generated pools and lanes are created through supported bpmn-js geometry, so participants and lanes render as expected
  • The admin user list reports each user's actual last login
  • The HIGH-severity findings of the July 2026 security audit are closed, including hardened Stripe webhook handling and tighter cross-tenant checks on federation collaboration endpoints
v2.0.2

Tidy-Up Layout, Save History & Workspace Invites

A gentle "Tidy up" layout mode with single-step undo, undo/redo buttons and a rolling save history in the editor, plus invites for not-yet-registered users and client-side diagram export.

Added

  • "Tidy up" layout mode in the editor — gentle align / grid-snap / edge-straighten that preserves your arrangement and reverts with a single undo
  • Undo / redo buttons in the editor toolbar
  • Rolling save history — up to 5 automatic snapshots per diagram (throttled ~1 per 5 min of editing, plus one before each auto-layout and AI change) with a preview-and-restore menu; restores are reversible
  • Full re-layout now offers a one-click Undo toast
  • Invite not-yet-registered users to a workspace by email — the invitation is applied when they sign up

Changed

  • The AI wizard's layout step now uses the gentle tidy-up instead of a full re-layout, keeping Wizzy's changes undoable

Fixed

  • Diagram SVG and PNG export renders in the browser and reports a failure instead of producing nothing
v2.0.1

Editor & Diagram-Listing Fixes

BPMN element coloring and single-lane naming restored, user-task forms save correctly, and the diagrams listing gains group-by, process metrics, and per-card forum links.

Added

  • Context-pad color picker for BPMN elements — recolor tasks, events, lanes and pools; colors persist through save/reload and XML export
  • Group-by toggle on the diagrams listing — folder, status, or folder-then-status with per-section counts and preserved sort order
  • Process-metrics card view on the diagrams listing — Lanes / Tasks / Paths / Systems / Maturity from the analytics snapshot, with a not-yet-analyzed state
  • Per-card forum link on the diagrams listing — open an existing linked thread or create-and-link a new discussion (workspace spaces only)
  • Lane Name field in the properties panel for single-lane pools

Fixed

  • BPMN element colouring runs through the bpmn-js renderer configuration, so themed defaults and custom colours both render
  • Form definitions built in the user-task dialog are persisted
  • A single-lane pool can be named from its properties panel, where the participant and lane label bands overlap on canvas
  • SVG and PNG export works for diagrams with no stored thumbnail
  • The landing page assistant answers from a maintained feature map rather than from the model's own assumptions
v2.0.0

The New Workspace Shell

A rebuilt workspace-centric UI with shell and switcher, Wizzy as public landing consultant with lead capture, a sharper six-pillar landing page, and Tresor as a new AI provider.

Added

  • Wizzy landing chatbot: a public, abuse-hardened AI consultant grounded in landing data, configurable from the admin panel
  • Lead capture with explicit consent — chat leads flow directly into the CRM
  • Configurable discovery-call booking URL replacing hardcoded links
  • Tresor as a stateless OpenAI-compatible AI provider

Changed

  • Navigation rebuilt workspace-centric: manage view as home base, persistent workspace shell, fast workspace switcher, and a personal shell
  • Landing page consolidated around the six CPG pillars with a unified section design and localized badges

Fixed

  • Protected pages render directly on a cold load — the auth guard runs server-side
v1.9.0

Stewarded Ecosystems

Workspaces for natural entities with legal personhood: ecosystem health index, care quests with human attestation, plus ISO/IEC 42001 compliance and server-side module enforcement.

Added

  • Stewarded Ecosystem workspace template: guardian/ranger/ecologist/volunteer roles, natural-asset registry, stewardship capability branch, and a ready-to-publish cleanup quest
  • Ecosystem Health Index with cockpit tab: score gauge, component breakdown, pain points, and a one-click create-quest path
  • Human-attested quests: volunteers accept and submit evidence, guardians approve or reject under a four-eyes rule, rewards issue automatically
  • Ecosystem-stewardship branch in the Global Technology Tree (charter, inventories, monitoring, cleanup, assessment, restoration)
  • ISO/IEC 42001 control library with Statement-of-Applicability management in settings, cockpit, and ISMS reviews
  • New landing page for stewarded ecosystems at /for-ecosystems, wired into the audience hub and signup attribution
  • CPG-cycle brand video featured in the landing hero

Changed

  • Logging in now lands in the workspace overview with your main workspace preselected
  • Paid add-on modules (CMDB, GTT, ESG, Academy Publisher) are now enforced server-side on every mutating API route
  • Quest publishing supports bounty-free internal workspace tasks with human attestation

Fixed

  • The onboarding tour keeps its place when you navigate back from the level page
  • Governance lane assignment offers an "(any)" placeholder for lanes without a specific person
v1.8.0

The Process Maturity Ladder

Processes now climb a unified L1–L7 maturity ladder — computed from workspace state and surfaced from the editor's readiness checks all the way to the workspace cockpit.

Added

  • Unified L1–L7 process maturity ladder, computed from workspace state and persisted per process (recomputed on freeze)
  • Pre-validation checks in the editor are grouped by maturity level, showing current standing and what unlocks the next rung
  • Maturity level surfaced across the workspace cockpit — governance overview and process rankings now span all seven rungs
  • New ProcesOS cycle glyph and animated loader; the CPG governance loop and L4–L7 rungs share one lifecycle visual
  • New gated Process Lifecycle knowledge-base article in the Academy
  • Landing pages: per-audience social-preview images, a default showcase image, and an audience navigation dropdown

Changed

  • Process level widened from 1–3 to the full L1–L7 maturity scale (backwards compatible — levels 1–3 keep their meaning)
  • The ProcesOS logo now links back to the homepage

Fixed

  • Login hero logo and headings are legible on the always-dark map background
  • Audience landing pages show their own per-audience image
v1.7.0

ERP Sync & Process Intelligence

ProcesOS now syncs live ERP data into process cost analytics, consolidates duplicate process roles, and reveals each process's most valuable and greenest path — under a refreshed brand identity.

Added

  • Odoo ERP integration: connect a workspace, map products to resource categories, and sync them with real costs into process analytics — plus inventory levels, KPIs, and a dry-run preview
  • Process-role consolidation: merge duplicate roles across diagrams using name, shared-responsibility, and member-name suggestions, with diagram sync-back and credential retitling
  • Events now carry directed value and resource flows, revealing each process's golden path (highest business value) and green path (lowest footprint)
  • CMDB sovereignty scan: a scan button auto-fills hosting, technology, and FAIR maturity details for IT-system artifacts
  • Refreshed ProcesOS brand identity — logo, favicon, brand color, and per-audience social-preview images, plus a new For Academia landing page
  • After login you now land in your main workspace cockpit instead of a blank editor
  • Per-audience conversion funnel for the landing pages

Changed

  • Paid add-on routes (Odoo, CMDB scan) are now enforced on the server, not just hidden in the UI

Fixed

  • Theme switching completes without a server error
  • The stakeholder validation dialog scrolls on small screens
  • The Academy back button returns you where you came from
  • XOR gateway branch probabilities are validated to sum to 100%
  • Unexpected server errors are caught and logged with a reference id you can quote in a bug report
  • Signup notifications describe the current activation flow
  • The user-task form editor renders correctly
v1.6.0

Governance Lifecycle & CMDB-Driven Process Analytics

Validation now gates AS-IS freeze behind an ISMS review + promote vote. Process Analytics' Systems tab realigned onto the real CMDB with FAIR/ISMS ratings.

Added

  • ISMS Compliance Review stage — when all lane-responsibles validate, the workspace's Head of Compliance gets an attestation task listing CMDB Systems + Data Objects used in the process, with FAIR / contract / GDPR gap indicators
  • Promote-to-AS-IS Vote stage — after ISMS attestation, a governance vote opens to lane responsibles + Head of Processes + Head of Compliance with configurable quorum and 7-day expiry
  • Governance dashboard surfacing — ISMS reviews and promote-vote tasks appear on the user dashboard, in the notification bell, and inline in the validation wizard
  • CMDB IT-System view in Process Analytics — Systems tab, Overview chart, Monthly cost table, and per-lane/per-path tags all driven by the real CMDB inventory with FAIR scores
  • Per-system Data Object detail — each CMDB system shows the Data Objects attached to its tasks with classification, personal-data flag, special categories, and legal basis
  • Shared freezeProcessAsIs() helper — atomic CAS-freeze + AS-IS version insert with rollback on failure

Changed

  • Stakeholder validation no longer auto-freezes the process — completion now opens the ISMS review stage instead. The validation-complete toast was repurposed to "Validation complete — ISMS compliance review opened"
  • Tailwind dark variant is now bound to the app's CSS-variable theme (light / dark / midnight / amethyst), not the OS prefers-color-scheme setting
  • Process Analytics snapshot stores real CMDB system names (instead of BPMN task-type pseudo-names) — improves AS-IS / TO-BE delta accuracy and SC4 setup-cost matching going forward
  • package.json now carries an explicit license field; README has a license section pointing at the LICENSE file

Fixed

  • Path analysis follows sequence flows only, so KPIs and path costs reflect the real process — data associations and message flows previously turned a two-branch diagram into 72 paths
  • The forum's process context resolves the CMDB systems and data objects a diagram actually references
  • Cockpit overview cards are legible on the light theme
  • Forum credential, tag and status badges hold their contrast on every theme
v1.5.4

Parallel Gateway KPI Correctness

Process-level KPIs no longer overcount shared tasks across paths produced by parallel gateways. Critical-path analysis replaces sum-of-durations for parallel branches.

Changed

  • Process-level KPIs (`totalCostPerRun`, `monthlyCost`, `totalEffortHours`) now use task-based aggregation with effective execution probability instead of summing over paths — eliminates the parallel-gateway overcount
  • Total process duration uses critical-path analysis (max across parallel branches) instead of summing across all paths
  • Scoped KPIs in `scope-analysis.ts` follow the same task-based pattern

Fixed

  • Process cost KPIs count each task once however many parallel paths contain it — an 11-task diagram with a single parallel fork previously reported roughly five times its cost per run
  • Process duration through a parallel fork uses concurrent-execution semantics, taking the longest branch instead of the sum of all branches
v1.5.3

Sovereignty Scanner, BYOK Wizard & Managed Agent Tooling

CMDB sovereignty scanner (SSRF-hardened), US CLOUD Act exposure in FAIR scoring, IRB FAIR matrix estimator, BYOK setup wizard, and BPMN tools on Anthropic managed agents.

Added

  • CMDB Sovereignty Scanner — server-side scan of DNS / HTTP security headers / SSL-TLS / WHOIS-ASN / technology detection for `it_system` artifacts (feature-gated, CMDB add-on)
  • Provider Knowledge Base — workspace-aware registry of hosting / CDN / email / analytics providers with jurisdiction (EU / US / other / unknown), seeded with ~50 hand-curated providers and growing automatically via scan-discovery
  • EU Sovereignty Score (0-100 + A-E grade) per artifact, with category-weighted breakdown (hosting 40%, email/auth 25%, CDN 15%, analytics 10%, other 10%)
  • US CLOUD Act exposure surfacing on scan results + propagation into FAIR scoring
  • IRB FAIR Maturity Matrix estimator from scan signals — full level descriptions per dimension stored alongside the score
  • SSRF-hardened scanner core — `validateScanTarget()` rejects private / loopback / link-local / cloud-metadata IP ranges with TOCTOU-safe IP pinning
  • BYOK Setup Wizard — guided multi-step flow (provider → test → model dropdown → managed-agent create → token usage)
  • Anthropic managed-agent BPMN tool integration — all 22 BPMN custom tools registered on the agent alongside `agent_toolset_20260401`
  • System-prompt teaching for the managed agent — explicit ProcesOS tool round-trip + result shapes baked into the agent body

Changed

  • BYOK key save now fails loudly when `WORKSPACE_KEY_ENCRYPTION_SECRET` is missing or misconfigured — silent encryption failure path closed
  • BYOK model-load now receives the API key explicitly from the connection test (no more post-save race against `getWorkspaceSecret()`)
  • Anthropic console link in BYOK setup updated to `platform.claude.com` (new Anthropic URL)

Fixed

  • Managed agent updates carry the version field the API requires for concurrency control, with one automatic retry on conflict
  • Empty star ratings in the validation wizard are visible against the background
v1.5.2

Wizard, Billing & Cost-Math Fixes + AI Model Discovery

Wizzy chat unbroken, Stripe boot-crash fixed, monthly cost math now probability-weighted, validation flow hardened. Plus AI model discovery dropdowns and password reveal on auth screens.

Added

  • Provider-agnostic model discovery in Admin AI Settings + Workspace BYOK Settings (dropdown after API-key entry, supports Anthropic / OpenAI / Mistral)
  • Anthropic skill / managed-agent listing in Admin AI Settings
  • One-click "Wizzy erzeugen" creates an Anthropic Skill agent idempotently (existing skill is reused, not duplicated)
  • Password reveal toggle (eye icon) on login, setup, and reset-password screens — preserves `autocomplete` for password-manager compatibility, aria-label i18n in en/de/es
  • Discovery-call CTA on the landing page linking to Proton Meet booking
  • Anthropic managed-agents Sessions API spike script (`scripts/spike-managed-sessions.ts`) — foundation for future Wizzy-on-Sessions integration

Changed

  • Stripe price-ID resolution is now lazy (resolved on first use, not at module import) — missing env vars no longer block server boot
  • Wizzy API now requires `spaceId` and resolves the AI provider per workspace (BYOK key + provider, then global, then env-var fallback)
  • Validation start now archives prior `validation_invite` notifications via `read: true` instead of deleting — audit trail preserved
  • Validation `complete` action performs the all-responsibles-validated check server-side and creates the AS-IS freeze atomically via CAS — no client-side race condition

Fixed

  • The AI assistant resolves the diagram's workspace on every message, so chat works from inside the editor
  • The server boots without a complete Stripe configuration — price identifiers resolve on first checkout instead of at startup
  • Monthly costs in the Analyze view weight each path by its probability, so a diagram with unconfigured gateways reports the real figure rather than up to three times it
  • The stakeholder validation Start button is disabled while a session is running, so a round cannot be started twice
  • A validation task on the dashboard opens the wizard directly
  • A process freezes to AS-IS automatically once every responsible stakeholder has validated
  • Auth screen sidebars fill their full height
v1.5.1

Legal, Security & Brand Hardening

TTDSG cookie consent, Terms of Service, GDPR-hardened legal pages, HTTP security headers, env-driven Stripe config, and a consistent ProcesOS brand across the app.

Added

  • TTDSG-compliant cookie consent banner gating non-essential storage
  • AGB / Terms of Service page
  • Expanded Impressum and Privacy Policy for GDPR compliance (controller, purposes, data-subject rights, processors)
  • Baseline HTTP security response headers on every response

Changed

  • Stripe price IDs are now environment-driven — test→live cutover is a pure env-var change with no code deploy or 500-error window
  • Checkout API now takes a tier key (`pro` / `team`) instead of a raw Stripe price ID; the server resolves it so price IDs never reach the client
  • App domain consolidated onto `procesos.techreform.de`; sitemap / robots.txt / llms.txt base URL is now env-driven via `APP_URL`
  • Product brand unified to ProcesOS across page titles, API docs, on-chain credential metadata, User-Agent headers, and UI text

Fixed

  • Menu sidebars fill their full height, with the navigation scrolling inside them
v1.5.0

Universal Quest System

Quests grow up: visibility/context discriminator, optional bounty, target-artifact rollup, auto-generation from CMDB gaps, and atomic credential + XP rewards on completion.

Added

  • `QuestPayload.visibility` (`internal` | `public`) and `QuestPayload.context` (`marketplace` | `workspace_task` | `maintenance`) as required discriminators
  • `QuestEvidenceKind` union — `validation_session`, `artifact_assessment_submitted`, `form_completed`, `manual_attestation`
  • `QuestTargetArtifact[]` rollup payload — multi-artifact, multi-kind quest targets supersede the single `requiredGttSlug`
  • Bounty is now optional — `workspace_task` and `maintenance` contexts accept `priceTier: 'free'` with no escrow checks
  • Bounty-free quests auto-close on activation (verified → closed) with two-hop `stateHistory` entries
  • **Artifact-completeness library** (`src/lib/server/artifact-completeness/`) — pure-function `computeCompleteness(artifact, domain)` with FAIR domain implementation for `it_system` artifacts
  • **Quest Generator engine** (`src/lib/server/quest-generator/`) — template-driven scanner with dedup, rate caps (50/workspace/day, 5/template/cycle), circuit-breaker, 24h scheduler, cockpit lazy-refresh
  • `questTemplates` + `questGenerationLog` collections with FAIR-on-`it_system` default template seeded per workspace
  • `quest_completion` credential type with `QuestCompletionMetadata` — atomic credential + Foundation Skill XP credit on quest completion
  • `QuestPayload.rewardSpec` — canonical reward configuration (credentialType + skill + xp) copied from generator templates at insert time
  • Compensating-rollback reward path — credential revoked if XP write fails; existing-credential pre-check prevents double-rewards on retry
  • Migrations 027 (visibility/context backfill), 028 (targetArtifacts from requiredGttSlug), 029 (questTemplates schema + indexes), 030 (seed default FAIR template into existing workspaces)

Changed

  • Quest publish validator now checks `visibility`, `context`, `evidenceSpec.kind`, `targetArtifacts` consistency, and optional `rewardSpec` (credentialType + xp + foundation-skill existence)
  • Q1g acceptance loop guards against non-`validation_session` evidence and issues reward *before* close-CAS for bounty-free path (failure leaves quest at `verified` for retry)
  • Marketplace QuestPublishPane hides bounty and party-kind fields for non-marketplace contexts
  • Workspace-creation routes (`/api/spaces/+server.ts`, `/api/spaces/from-template/+server.ts`) seed the default FAIR quest template
  • Cockpit page-load triggers a lazy quest-refresh (debounced, fire-and-forget with `.catch()` to prevent SvelteKit unhandled-rejection crashes)
v1.4.0

Maturity Epic Sealed + Cockpit Systems Tab

Maturity Epic #104 sealed: all 12 profile-types covered, age-tier-cap, admin-tunable weights, history sparklines. Plus Cockpit Systems tab with FAIR/ISMS scoring and 3-tier cost analysis.

Added

  • Maturity profiles for the 7 remaining ecosystem-roles (`district`, `association`, `organisation`, `transform_consultant`, `agency`, `government`, `solution_provider`) — no more silent business-fallback
  • Workspace-age tier cap (silver under 3mo, gold under 12mo, unlimited ≥12mo) with would-be-tier display in Cockpit Maturity hero
  • Admin-tunable maturity weights via Cybernetic Core — per-profile, per-dimension override with 1.0 sum validation
  • `Admin → Cybernetic Core → Weights` tab with profile selector + dimension table + sum-indicator + reset
  • Maturity history snapshots (`workspaceMaturitySnapshots` collection) with daily-deduped 4-hour scheduler over active workspaces
  • Inline sparkline trend in Cockpit Maturity hero — per-dimension and overall score
  • `GET /api/spaces/[id]/maturity/history?days=N` endpoint with 5-min client-side TTL cache
  • Cockpit **Systems** tab — CMDB-sourced view of every IT system with FAIR / ISMS / contract / ESG / AI metadata
  • FAIR dimension visualization (5-segment horizontal bars per dimension, color-coded by level)
  • Framework filter toggles on Systems tab (GDPR / NIS2 / DORA / ISO27001) with OR-semantics across multi-select
  • Sort options on Systems tab — sovereignty / FAIR-score / name / cost / criticality, with rated-first / unrated-last for sovereignty + FAIR
  • Header counter badges on Systems tab — `X low sovereignty · Y not assessed` for at-a-glance pain-point context
  • 3-tier cost analysis with execution multipliers and time-window calculations (path / process / global metrics)
  • `src/lib/bpmn/cost-math.ts` — pure-math helpers extracted as testable module (21 unit tests)

Changed

  • Maturity engine `getProfileFor` async-resolves admin overrides via derived-config; sync `getDefaultProfile` exposed for tests; `getProfileFor` kept as `@deprecated` alias
  • Maturity engine `getProfileFor()` default branch now only catches `null` / `undefined` ecosystem-role; all 12 roles resolve to dedicated profiles

Fixed

  • The workspace age notice uses correct plurals in all three languages
v1.3.0

Treasury, Maturity, Audience Pages

Workspace Treasury sealed end-to-end. Maturity engine + 12 dimensions visible in Cockpit. Six audience-targeted landing pages live. Plus dashboard worldmap, listing hardening, cockpit hub.

Added

  • Workspace Treasury — non-custodial declared-wallet, CIP-30 ownership proof, Blockfrost balance + 6h cache, rotate / unset with open-escrow guard, audit log UI, obligations chip
  • Workspace-funded quests end-to-end — Head-of-Finance publishes quests from the workspace treasury; fulfillers offer via per-user `/api/user/me/treasuries`
  • Workspace Maturity engine with 12 dimensions across 5 ecosystem-role profiles (business / community / village / institution / university)
  • Cockpit Maturity tab — score ring, tier badge (bronze→diamond), dimension sub-score grid weighted by `effectiveWeight`, per-role rating breakdown, embedded Skill + GTT-capability dashboards
  • Audience-targeted landing pages: `/for-business`, `/for-enterprise`, `/for-communities`, `/for-institutions`, `/for-government`, `/for-solution-providers`
  • Dashboard worldmap with workspace + federation-peer markers, Nominatim-backed address search
  • Cockpit Übersicht redesign — summary-card hub with deep-links to detail tabs
  • GET `/api/spaces/[id]/maturity` (auth-gated, 5-min client-side TTL cache)
  • GET `/api/spaces/[id]/treasury/obligations` (any workspace member, returns `{ count, totalLovelace }` for live workspace-funded quests)
  • GET `/api/user/me/treasuries` (returns spaces where current user holds Head-of-Finance + a verified treasury)
  • ADR-0001: rating-infrastructure decision for the role-experience-variance dimension

Changed

  • `/api/marketplace/listings` GET returns tiered responses based on caller auth (anonymous → identity-stripped, authed non-member → `kind` only, creator-space member → full payload). Single batched membership lookup avoids N+1.
  • Treasury DELETE handler shares the same `getOpenObligationsSum` helper as the new obligations endpoint — single source of truth for open-escrow precondition.
  • Marketplace `?creator=<spaceId>` filter param threads through to listings GET — used by the worldmap quest-badge deep-link.
v1.2.0

Quest Marketplace & Compliance Cockpit

Demand-side marketplace: workspaces publish capability bounties in ADA, fulfillers deliver via blueprints, acceptance auto-triggers on GTT activation. Plus compliance cockpit + voting view.

Added

  • Quest Marketplace (Epic #95): publish a capability-activation bounty with ADA locked in Cardano escrow; fulfillers offer, deliver, and get auto-accepted on GTT activation
  • Quest discovery tab at /marketplace?type=quest with bounty/deadline/urgency card, scope badge (U→U / U→W / W→U / W→W), archive toggle
  • Inline OfferToFulfilDialog — no detail page, one-click from frontier-pill or marketplace-card to offer-submit
  • GTT Frontier bounty chip: any unblockable capability with an open quest shows an accent ₳-pill that deep-links into the offer flow
  • Automatic quest acceptance loop: when the required GTT node flips to `active`, matching open quests auto-advance to `verified` with forensic evidence (non-fatal sweep, CAS-guarded against double-writes)
  • quest_fulfilment credential minted on offer-accept; space-scoped lookup so it survives individual member departures
  • Workspace Cockpit Compliance Tab: cross-diagram aggregation with min() semantics, per-framework/jurisdiction heatmap, pain-point hero card, gap click-through
  • Multi-dimensional Delta View (BPMN + cost + resource + GTT + compliance) extracted into its own reusable component
  • Voter View for change proposals with credential-gated voting + weighted tallies + on-acceptance bounty release
  • Workspace Treasury backend primitive: non-custodial CIP-30 signData ownership proof, Mongo-backed nonce store with atomic consume-once replay protection, rotate/unset endpoints that respect open-escrow immutability (UI lands v1.3.0)

Changed

  • Validated AS-IS BPMN diagrams are now immutable — PUT handler returns 409 `edits_disabled_after_validation` once a governance validation session has locked the process; proposals must flow through the voting view
  • Marketplace browse query filters out expired quests (no auto-cron for published → expired) and self-authored quests (no circular self-nudge)
  • Compliance analysis moved server-side via bpmn-moddle, with a 5-minute client-side cache on the aggregated workspace compliance store
  • QuestPayload's `questParties.quester` is the single source of truth for funder identity — open-escrow states (`published`, `in_progress`, `verified`, ...) exported as `OPEN_QUEST_STATES` for cross-module reuse

Removed

  • User-level subscription model — licensing is workspace-only now; personal spaces cascade-delete with their owner
  • Hard-coded `fundedFromSpaceId` / `fundedFromTreasuryAddr` escrow snapshots — superseded by QuestPayload's discriminated-union funder encoding + AS-IS payload immutability
v1.1.0

Marketplace & Sovereignty

Operating Layer epic complete: unified paid-escrow marketplace across all item types, capability activation loop, sovereign workspace identity via KERI AID, Forms linked to DataObjects.

Added

  • Unified paid-escrow marketplace across blueprints, GTT capabilities, artifacts and workspace templates
  • PublishListingDialog: five-step type-aware publish wizard with early-draft persistence
  • Creator Analytics dashboard at /marketplace/creator with totals, 30-day trend chart, acceptance rate
  • Blueprint.producesGttSlug hook: run-context validation auto-activates the referenced GTT capability
  • Editor action "Solution from selected artifacts" — promote artifact-bearing tasks into a workspace solution node
  • Durable audit log for admin-triggered GTT frontier snapshots, surfaced in the admin debug view
  • KERI AID support via KERIA integration — per-workspace identity with on-chain registration (Cardano metadata label 7743)
  • Public workspace profile at /w/[slug] with tokenize.it invest-widget integration
  • Head-of-Finance role with treasury:manage permission scope
  • Forms now link to DataObjects; AI heuristic infers GDPR categories from field keys and labels
  • DataObject settings: linked-forms section + read-only form preview on the GDPR tab
  • 30-day Pro trial on every new workspace (pre-Stripe launch)
  • Vitest infrastructure as first repo-wide unit-test runner
  • In-app Bug Report dialog creates GitHub issues via server-side PAT — no repo access needed for users

Changed

  • WorkspaceCreateModal rewritten on top of a live marketplace picker — any published workspace_template listing appears automatically
  • FilterBar sector dropdown expanded to full FoundationSector enum; gains optional priceTier opt-out
  • MarketplacePublishSettings now dispatches to the unified PublishListingDialog with preset type
  • Tier-gate removed from workspace creation — every new workspace starts Pro-trialing
  • Frontier admin UI renders a compact timeline table instead of prose snapshot metadata
  • UserTask properties panel always offers a Create-form button, not just an Edit button for already-linked forms

Fixed

  • Inline label editing is legible on every theme, caret included
  • Spell-check underlines no longer appear on BPMN element labels
  • The "Add form" dialog opens when you ask for it, rather than on every load of a diagram that already has forms
  • The form editor renders on the first open of the dialog
  • The lane roles tab loads instead of spinning indefinitely
  • Docker images build cleanly with the credential signing dependencies
  • Form save errors surface as a notification, and the list refreshes once a save succeeds
  • The forum credential badge renders for the Head of Finance role
  • The editor route renders server-side without error

Removed

  • Legacy /api/marketplace/templates/* endpoint family (use /api/marketplace/listings/* instead)
  • PublishWorkspaceDialog component (superseded by PublishListingDialog)
  • Per-creation module-checkbox add-ins on WorkspaceCreateModal (Launch-Trial activates all modules)
  • Recommendation-tier match-badge UI (tier-gate no longer relevant with Launch-Trial)
v1.0.2

Operating Layer: Artifacts, Unified Marketplace, Contextual GTT

Operating Layer launch: unified marketplace, CMDB artifacts, contextual GTT frontier, and artifact/capability references directly on BPMN tasks.

Added

  • Artifact/Capability two-layer data model with cmdbArtifacts collection and artifact-bound GTT nodes
  • Unified marketplace with four item types (blueprint, gtt_node, artifact, workspace_template) and faceted filters
  • Workspace Capability Tree seeded per space (procesos-workspace, workspace_core and sovereign/intelligence/operations/interop layers)
  • Contextual GTT Frontier panel surfacing the next three recommended nodes with explainable rationales
  • Adoption gates on GTT nodes: required credentials, workspace skill aggregates, and blueprint listings
  • bpb:GttNodeReference as a first-class BPMN palette element with status-aware renderer
  • artifactRefs and gttNodeRefs on BPMN flow nodes and lanes with a dedicated properties-panel section
  • Artifact editor UI with type-specific fields for physical, digital asset, sensor, document and integration artifacts
  • Free-adopt endpoint for marketplace listings, distinct from paid purchase flow
  • Admin debug view for the GTT Frontier with snapshot, recompute and heatmap tabs
  • Four-act scripted story and milestone banner on the login map

Changed

  • Settings tab 'CMDB / IT-Systems' renamed to 'Artifacts', with IT-systems now one category among many
  • Workspace template selection is sourced from marketplace listings instead of hardcoded seeds
  • Governance validator now checks referenced artifacts are deployed and capabilities active
  • Marketplace listings support structured jurisdictions and compliance frameworks instead of free-text tags
  • GTT template titles, descriptions and rationales are fully i18n-keyed with fallbacks

Fixed

  • The editor page loads reliably again
  • Production images build within the available memory budget — API documentation generation is skipped in production builds
v1.0.1

Marketplace, Process Roles & Release Communication

Cardano-secured workspace template marketplace, BPMN lanes promoted to workspace-wide process roles, a public changelog, and various UX improvements.

Added

  • Cardano smart-contract-secured workspace template marketplace (Publish → Buy → Instantiate → Validate → Release/Refund)
  • Anonymization pipeline for template export with PII detection and deterministic role-placeholder remapping
  • BPMN lanes as first-class workspace process roles with credential gates and default cost rates
  • Public /changelog page with release timeline and empty state
  • Admin release management panel (CRUD, draft/published workflow)
  • "What's New" banner in the layout, tied to the latest published release version
  • App version and changelog link in the footer
  • Forgot password flow — password reset directly from the login screen
  • Form.js integration for UserTask form definitions in the editor
  • Self-sustaining village demo workspace template (~125 seed documents)
  • World events stream and resource flow panel on the login screen
  • ProcesOS brand definition with tagline typewriter in the hero section
  • /marketplace/how-it-works transparency page with links to validator and signer source

Changed

  • Lane roles auto-register workspace-wide on diagram save
  • Credential gate: hard for responsible, soft warning for contributor/viewer
  • Archived roles auto-reactivate when the same lane name reappears
  • Cardano NFT portfolio now queries by stake address for full HD wallet coverage

Fixed

  • Cardano network selection is explicit — an unconfigured network is refused rather than quietly falling back to preprod
  • Login map connection lines are subtler, and quest bubbles stay inside the viewport
v1.0.0

ProcesOS — The Process Operating System

Initial release of ProcesOS: a multi-tenant process management platform combining BPMN 2.0, DMN decision tables, AI assistance, process governance, and capability planning into a composable process execution language for organizations.

Added

  • BPMN 2.0 editor with bpmn-js — full diagram modeling, auto-save, translation overlays, auto-layout (ELK.js)
  • AI Wizard Assistant — multi-turn AI chat for diagram creation and modification (Anthropic, OpenAI, Mistral providers)
  • DMN Decision Tables — dmn-js integration linking BusinessRuleTask elements to reusable decision table definitions
  • Multi-tenant workspaces with subscription tiers (free, pro, team, enterprise) and feature gates
  • Process Governance — AS-IS/SHOULD-BE versioning, stakeholder validation wizard, change voting
  • Global Technology Tree (GTT) — 5-element capability planning (Fire/Water/Earth/Air/Aether) with status machine and unlock cascades
  • CMDB — IT systems registry with cost tracking, license management, FAIR assessment, and ESG data
  • Data Objects library with GDPR classification (personal data flags, legal basis, retention periods)
  • Resource tracking with carbon footprint analysis per process path
  • Cost & ROI analysis — per-task cost modeling, lane costs, process-level analytics with currency support
  • Compliance analysis — regulatory gap detection, DPIA/GDPR fields, EU AI Act properties (Art. 14, 26)
  • FAIR Score assessment for IT systems (Transparency, Accountability, Interoperability, Data Sovereignty, Human Oversight, Local Adaptability)
  • Discussion forum with credential-gated access, diagram linking, AI insights, and change proposals
  • Gamification — XP-based skill system with bronze-to-diamond tiers, foundation skill graph, and mastery levels
  • Cardano/Web3 — wallet connection, credential minting, NFT portfolio, Andamio integration
  • Stripe billing — subscription management with webhooks
  • Cross-instance federation — Cardano registry, Ed25519 signing, public surfaces, collaboration references
  • Cybernetic Core — server-side configuration engine with pluggable engines (feature gates, pricing, landing page)
  • CPG Demo Workspace (Baeckerei Schmidt) — 4 processes, 8 IT systems, 6 data objects, 7 resources, full governance setup
  • Internationalization — English, German, Spanish (32 namespaces via Paraglide/inlang)
  • Better-auth authentication with 33 social providers, organization-based permissions, API key support
  • Interactive guided tours (Shepherd.js) for onboarding