docs: organize cross-product documentation
Dependency Audit / dependency-audit (push) Successful in 1m46s
Deployment Installer / deployment-installer (push) Successful in 9s
Security Audit / security-audit (push) Successful in 11m48s

This commit is contained in:
2026-08-17 16:52:51 +02:00
parent 209a43592f
commit c66e1b768d
46 changed files with 384 additions and 271 deletions
+198
View File
@@ -0,0 +1,198 @@
# GovOPlaN Platform Core Ideas
## Purpose
GovOPlaN is an institutional governance and operations layer. Its central
promise is:
> Model the institution, orchestrate its work, connect its systems, and
> preserve why and under whose authority it acted.
The platform should let people complete a real task without understanding its
repository or module graph. It should let institutions retain control over
their data, procedures, providers, and deployment while still sharing
interoperable definitions and evidence.
This document is the stable summary of the ideas that every product package,
module, interface, and integration must preserve. Current implementation state
lives in [Strategy Status](STRATEGY_STATUS.md).
## Ten Core Ideas
### 1. Institutional context before application context
Work happens for a tenant, institution, organizational unit, function,
mandate, jurisdiction, service, case, and represented party. The real actor
and represented capacity remain distinct. Application permissions alone do not
prove institutional competence.
### 2. Governance is executable
Policy is not explanatory prose around an operation. Consequential actions
must expose applicable rules, authority, purpose, expected effects, review
requirements, recovery behavior, and evidence. Inheritance may tighten a rule
but must not silently loosen an upstream constraint.
### 3. Time has two independent meanings
Valid time answers when a fact applied. Recorded time answers what the system
knew at a point in history. Historical browsing changes the business-data
projection, never the current authorization context. Corrections and
supersession remain visible rather than rewriting history.
### 4. One context, many owners
Cases, tasks, decisions, records, messages, files, appointments, reports, and
external objects remain owned by their domain modules or source systems. Stable
references create one navigable context without a universal copied master
record or cross-module table access.
### 5. Native and connected operation are peers
For every integration, GovOPlaN states whether it is authoritative, mirrors an
external source, synchronizes governed fields, adds a governance overlay, or
keeps a link only. An external system can be used today and replaced later
without losing provenance or institutional control.
### 6. Human work is a first-class system object
An intake becomes owned, reviewable work. A person can see the current context,
next responsible action, reason, deadline, consequence, and completion
evidence. Workflow Engine coordinates machine and human transitions; focused
views guide people through the relevant platform surfaces.
Tasks owns explicit work items and the unified work inbox. Workflow Engine owns
process execution and resumable handoffs. Notifications attract attention, and
domain modules retain their business objects. These boundaries prevent an
inbox, workflow, or notification from becoming a second copy of institutional
state.
### 7. Views reduce complexity without changing authority
The interface is a task- and role-sensitive projection of installed
capabilities. Views, dashboards, search, documentation, and workflow-guided
surfaces may hide irrelevant functions, but they never grant access. Users can
escape a focused mode when policy permits and can always understand why
something is unavailable.
Configurable product areas organize authorized capabilities around work,
services, records, communication, meetings, data and institutional
responsibility. The optional Quick Access rail presents task-local Work,
Calendar, Messages and Files contributions without merging their owners or
turning presentation settings into permissions.
### 8. Evidence and recovery are part of the operation
Intent, exact input versions, approvals, external effects, receipts,
outcome-unknown states, reconciliation, corrections, retention, and recovery
belong to one evidence chain. A retry must be idempotent; rollback claims must
distinguish reversible local state from effects already observed elsewhere.
### 9. Inclusion is multi-channel, not portal-only
Public portal, postbox, mail, telephone, paper, in-person assistance, APIs, and
external systems are channels around the same governed work. Assisted entry
records who entered information, for whom, from which source, with which
attestation, and how the affected person receives a usable receipt and
correction path.
Responsive, mobile, desktop, and embedded launch surfaces are additional ways
to enter the same governed context, not separate products with weaker authority
or evidence. Common task-local actions may open in bounded overlays while their
owning modules retain validation, policy, and persistence.
### 10. Successful configurations are portable products
Modules are ingredients. A usable product is a signed configuration package
with terminology, forms, policies, workflows, views, reports, provider
profiles, documentation, migration rules, and evidence. Institutions derive
local packages without forking code or weakening inherited constraints.
## Platform Planes
The planes below are ownership lenses, not navigation groups or mandatory
deployment tiers.
| Plane | Responsibility |
| --- | --- |
| Experience | Shell, views, dashboard, search, help, accessibility, and task-focused composition |
| Participation and channels | Portal, postbox, mail, campaigns, calendar, scheduling, consultation, and assisted channels |
| Human work and procedure | Services, forms/runtime, cases, tasks, approvals, workflow execution, and domain procedures |
| Content, records, and evidence | Files, templates, DMS, eAkte/records, audit, reporting, transparency, and publication |
| Institutional governance | Identity, access, tenancy, organizations, functions, mandates, policy, trust, and formal decisions |
| Data and integration | Connectors, datasources, dataflow, search, external references, provider health, and reconciliation |
| Runtime and assurance | Module composition, operations, deployment, recovery, security evidence, and signed packages |
Collected product ideas and normalized actor outcomes are preserved in the
[Product Input Register](PRODUCT_INPUT_REGISTER.md). They enter implementation
only through a named journey, package, or explicit discovery issue.
## Canonical Distinctions
The platform must not collapse these pairs:
- identity vs account vs represented capacity;
- role/permission vs function/mandate/competence;
- valid time vs recorded time;
- purpose for use vs general technical access;
- document content vs managed file bytes vs institutional record;
- task vs workflow definition vs workflow instance;
- approval vs formal decision;
- message intent vs transport delivery vs recipient acknowledgement;
- source authority vs connector maturity;
- current state vs historical evidence;
- correction/compensation vs erasure of an observed effect;
- a module boundary vs a user-visible product boundary.
## Product Experience Rule
The normal user interface speaks in services, work, records, messages,
meetings, decisions, and outcomes. Module names, provider IDs, capability names,
package coordinates, and schema details are technical provenance. They are
visible to administrators and in expandable diagnostics, but they are not the
primary information architecture for ordinary work.
The complete permission-derived tool catalogue remains deliberately available
to power users. Product areas and Quick Access provide sensible system and
tenant defaults plus governed user personalization; they do not make familiar
tools harder to reach merely to conceal modular implementation.
## Maturity Rule
A repository, route, model, or unit test does not make a capability complete.
Claims advance only with evidence appropriate to the claim:
1. `scaffold`: boundary and documentation exist;
2. `vertical_slice`: useful behavior has focused tests;
3. `reference_ready`: an end-to-end reference journey passed target,
accessibility, privacy, security, operations, and recovery evidence;
4. `supported`: upgrades, interoperability, support procedures, and release
guarantees are defined;
5. `lts`: compatibility and maintenance windows are contractual.
## Deliberate Non-Goals
GovOPlaN does not aim to:
- replace every specialist system, ERP, DMS, groupware, or data tool;
- make one database authoritative for every connected fact;
- expose every installed capability to every person;
- infer authority from organizational membership alone;
- make historical browsing weaken current security;
- treat AI output as an unaccountable institutional decision;
- create a repository for every noun in the information model;
- claim production maturity from local development evidence.
## Decision Test
A proposed feature fits the platform when it improves at least one real
institutional journey and can answer:
1. Who owns the object and source of truth?
2. In which institutional and temporal context does it apply?
3. For which declared purpose may it be used?
4. Which policy and authority permit the action?
5. What effect, evidence, retention, and recovery behavior result?
6. How can it operate with an external owner without losing autonomy?
7. How will a person discover and complete it without learning the module
graph?
+203
View File
@@ -0,0 +1,203 @@
# GovOPlaN Product Input Register
## Purpose
This document preserves and normalizes product ideas and user-story notes that
inform GovOPlaN without turning a private note file into a second backlog.
Gitea issues remain the source of live work state; the stable platform direction
remains in [Platform Core Ideas](PLATFORM_CORE_IDEAS.md), the
[Connected Governance Platform Roadmap](reference/CONNECTED_GOVERNANCE_PLATFORM_ROADMAP.md),
and the [Reference Journey Program](REFERENCE_JOURNEY_PROGRAM.md).
The register was reconciled on 2026-08-06 from:
- `/mnt/DATA/Nextcloud/ADD ideas UG/Products/govoplan/ideas.md`;
- `/mnt/DATA/Nextcloud/ADD ideas UG/Products/govoplan/user_stories.txt`.
The source notes remain useful as the original capture. This maintained version
uses consistent terminology, makes ownership explicit, and records where an
idea enters the product program.
## Product Themes
### Operable and scalable installation
An operator should be able to install, update, reconfigure, scale, back up,
restore, pause, and retire GovOPlaN through one explainable control plane.
Existing infrastructure may be reused or managed components may be provisioned.
The WebUI and CLI must invoke the same governed operations, show the planned and
completed effects, preserve recovery evidence, and never claim rollback for an
external effect that cannot actually be reversed.
This theme is owned by Core, Admin, Ops, Policy, Files, and the signed product
package. It is tracked primarily by GovOPlaN #13 and the production evidence
issues. It advances in parallel with, but does not replace, actor-facing
reference journeys.
### Focused, consistent work
People should see the work and tools relevant to the current task, not the
installed module graph. Views may be defined by administrators, groups, or
users within policy. Workflow instances may pin a governed View. Contextual
help, predictable action placement, consistent central components, visible
intermediate results, and plain institutional terminology are product
requirements.
Small task-local actions such as writing a Mail or Postbox message, completing
a Template, or manipulating Files should be launchable without abandoning the
current context. These actions remain owned by their modules and use bounded
overlays or workspaces; the shell supplies discovery and return context rather
than reimplementing them.
The accepted first presentation is the optional, configurable Quick Access
rail: Work, Calendar, Messages and Files. Messages may compose Mail, Postbox
and future chat contributions while preserving their separate authority and
channel semantics. System and tenant administrators govern availability and
forced entries; users select categories and ordering within those ceilings.
Modules register typed contributions through Core and continue to work when
Quick Access is absent.
This theme is owned by Core experience contracts, Views, Dashboard, Tasks,
Workflow Engine, Quick Access, Docs, and the contributing feature modules. The
first proof is the resumable service-to-decision/eAkte journey in GovOPlaN #42.
### Governed human work
An intake or event becomes owned work with a responsible actor or function,
priority, deadline, current action, consequence, source context, and completion
evidence. Tasks owns explicit work items and the unified work inbox. Workflow
Engine owns process execution, waits, retries, and handoffs. Domain modules own
the business objects and commands. Notifications attract attention but do not
replace durable work state.
This distinction applies to service requests, technical support, approvals,
data reconciliation, campaigns, meetings, decisions, records, and failed
automation. It is the immediate shared implementation priority because users
must be able to leave work and resume it safely.
### Institutional responsibility and workforce context
Organization units, functions, mandates, assignments, delegations, and acting
context determine institutional responsibility. Presence, absence, illness,
availability, and similar status are effective-dated operational facts used to
route work, suppress or redirect notifications, explain planning, and trigger
policy. They are not merely profile decorations and they do not replace the IDM
lifecycle status of an identity or account.
Time recording, absence management, sickness reporting, return-to-work
management, and applicant management form a possible workforce package. The
first implementation must be driven by a real journey and legal/privacy
profile; no new module boundary is implied solely by this register.
### Integration-first and provider-neutral operation
GovOPlaN should integrate tightly with software already used by an institution
and offer native alternatives only where that produces a better governed
outcome. Core-mediated provider contracts expose stable, vendor-neutral
capabilities; adapters encapsulate specific products. Authority, synchronized
fields, conflict behavior, health, credential custody, provenance, and
retirement must be explicit.
The LBV Baden-Wuerttemberg idea is retained as a candidate workforce/payroll
integration profile and as a test of provider-neutral contracts. Desktop and
groupware integration for Microsoft Office, Outlook, LibreOffice, Thunderbird,
file managers, Windows, Unix, and macOS should use standards, deep links,
protocol handlers, synchronization, and governed connectors before custom
desktop software is introduced.
### Inclusive channels and device surfaces
Portal, Postbox, Mail, telephone, paper, in-person assistance, API, desktop,
and mobile are channels around the same governed work. A responsive or native
mobile surface must not create a second authority or data model. Assisted work
records representation, source, attestation, receipt, correction, and delivery
choice. People may opt into permitted distribution channels while policy keeps
mandatory channels and legal delivery requirements explicit.
Video meetings, chat, instant messaging, and forums are retained as governed
collaboration-channel candidates. The default direction is integration with an
established provider through typed message, meeting, participant, evidence, and
retention contracts before building another communications stack.
### Meetings, deliberation, decisions, and voting
An institutional meeting spans scheduling, participants and mandates,
documents, agenda, discussion, formal motions, votes, decisions, minutes,
follow-up work, publication, and eligible expense settlement. Committee owns
the meeting and deliberation semantics while Calendar, Scheduling, Files,
Templates, Decisions, Tasks, Reporting, Ledger, and Voting contribute optional
capabilities.
Voting requiring certified assurance remains a provider program. POLYAS is the
first external profile; a native provider may progress only through the
controlled assurance and certification program already tracked in Voting.
### Controlled data work and understandable reporting
People should manipulate data through immutable inputs, previewed operations,
intermediate materializations, reversible definition changes, durable review
decisions, quality rules, and complete lineage. Reports expose their definitions
and source revisions so controllers can understand and change how a result is
produced. Technical support may package controlled workflows that let
non-technical users safely operate otherwise hidden data.
The monthly-data journey is the first proof. Sanctions screening follows on the
same source, snapshot, transformation, review, reporting, workflow, and
delivery contracts.
### Institutional memory and consequence
Decisions should be prepared, discussed, made, communicated, implemented, and
filed with their authority and consequences visible. A record/eAkte provides
the familiar administrative context across exact source revisions without
copying ownership from Cases, Decisions, Files, Forms, Campaign, Postbox, or
other modules. The institutional digital twin may later use governed
projections to model and simulate organizational change, but simulation output
never becomes authority without an explicit adoption decision.
## Normalized Story Catalogue
The following catalogue preserves the intent of the collected notes. It is an
orientation index, not a completion checklist.
| Actor and desired outcome | Product owner or composition | First proof |
| --- | --- | --- |
| Operator installs, updates, scales, backs up, restores, and rolls back through one explainable workflow | Core, Admin, Ops, signed package | GovOPlaN #13 and target-evidence lane |
| System and tenant module administrators govern module availability and lifecycle | Core, Admin, Policy, Tenancy | Module entitlement and lifecycle composition |
| User works in a decluttered, consistent and task-sensitive interface | Views, Core, Dashboard, Workflow, Docs | Service-to-decision workspace |
| Policy maker defines inherited, explainable and enforced rules | Policy plus every consequential owner | Information-governance adoption gate |
| Controller and auditor reconstruct results, rules, evidence and correction paths | Audit, Reporting, Records, Dataflow | Monthly-data and eAkte journeys |
| User sees institutional terminology, current progress, intermediate results and consequences | Domain owner, Tasks, Workflow, Views | All reference journey acceptance tests |
| Voting body and voter obtain independently assured democratic voting | Voting, Committee, Identity Trust, Encryption | POLYAS profile and controlled native-provider program |
| Support staff packages safe guided manipulation of hidden data | Workflow, Dataflow, Tasks, Views | Monthly reconciliation workflow |
| Data worker performs controlled, understandable and recoverable transformations | Datasources, Connectors, Dataflow, Reporting | GovOPlaN #8 |
| Decision maker prepares, deliberates, decides, records and follows consequences | Committee, Decisions, Tasks, Records, Reporting | Service-to-decision journey |
| Management delegates responsibility and receives governed activity reports | Organizations, IDM, Access, Policy, Reporting | Function-bound Postbox and work inbox |
| Institution models and simulates organizational change | Organizations, Policy, Dataflow, Reporting, Digital Twin | Later governed digital-twin package |
| Sender distributes generated files to functions without knowing incumbents | Campaign, Distribution Lists, Postbox, Organizations, IDM | Governed communication package |
| Function holder receives current and policy-selected historical work and information | IDM, Access, Postbox, Tasks, Records | Postbox reassignment/history tests |
| Administrative worker accesses one familiar eAkte context across exact owned objects | Records and record-source providers | GovOPlaN #42 and Records #8 |
| User invokes common message, template and file actions without leaving the current task | Core shell, Views, Workflow and contributing modules | Task-local action contract and service workspace |
## Idea Preservation Map
| Original idea cluster | Preserved direction |
| --- | --- |
| Time recording, absence, sickness, reintegration, applicant management | Governed workforce-context journey; effective-dated status and privacy profile before module expansion |
| LBV BW interface | Candidate provider-neutral workforce/payroll connector profile |
| Abstract interfaces | Versioned Core contracts with product adapters and explicit source authority |
| Desktop and groupware integration | Standards, connectors, deep launch and synchronization before custom clients |
| GovOPlaN app/mobile-first pages | Responsive shared semantics; native shell only when a proven journey needs device capabilities |
| Video, chat, instant messaging and forum | Optional governed collaboration providers with retention/evidence contracts |
| Somacos Session-style meeting management | Committee-led meeting composition across Calendar, Files, Decisions, Templates, Tasks, Reporting and Ledger |
| Stronger software integration | Integration-first roadmap rule and first full external product connectors |
## Maintenance
When a source idea becomes actionable:
1. link it to a named reference journey or explicit discovery issue;
2. identify the owning module and external authority;
3. create or update the Gitea issue with acceptance criteria;
4. keep live status out of this document;
5. update this register only when the durable interpretation changes.
+517
View File
@@ -0,0 +1,517 @@
# GovOPlaN Reference Journey Program
## Status
This is the selected product-development sequence, originally chosen on
2026-07-21 and reconciled with the implemented platform on 2026-07-31. It turns
the long-term connected-platform roadmap into five demonstrable journeys.
Workflow Engine and the optional Workflow editor now exist, but they are used
by a stage only when its package explicitly composes and proves them; Workflow
is not an automatic dependency of every journey.
The institutional semantics and source-authority model applied to these stages
are defined in the
[Institutional Governance Target Architecture](../architecture/INSTITUTIONAL_GOVERNANCE_TARGET_ARCHITECTURE.md).
The stages are ordered, but they are not monolithic releases. Each stage is
delivered as small, reviewable, green increments and is complete only when its
user journey, failure behavior, documentation, and operator evidence work in a
pinned composition.
## 2026 outcome reset
Repository completion is not product completion. From 2026-08-05 onward, work
is accepted primarily through three maintained real-life journeys:
1. **Governed communication:** select accountable recipients, prepare content
and attachments, approve, deliver through Mail and/or a function-bound
Postbox, reconcile uncertain outcomes, and file the evidence.
2. **Inclusive service-to-decision:** accept a request through a digital or
assisted channel, establish identity and purpose, guide the case through
human and automatic work, decide, notify, and file the resulting eAkte.
3. **Monthly data and sanctions:** acquire immutable source snapshots, validate
and reconcile them interactively, preserve decisions and lineage, produce
reports and files, and deliver the accepted result through Campaign.
The staged program below remains the architectural build order. These journeys
are the acceptance lens across those stages. Every significant feature should
identify the journey it improves, or provide security, operability, recovery,
accessibility, or usability evidence that those journeys require. Work that
does neither stays in the backlog until a concrete consumer exists.
The Records vertical now supplies the journey's native file plan, immutable
record and item revisions, chronology, close/reopen, retention calculation,
holds, appraisal, independent disposition approval, recovery-ledger evidence,
and archive-neutral package simulation. Forms Runtime, Cases, and Decisions
expose exact, permission-rechecked source revisions for explicit filing, and
all three contribute metadata-only native Search projections that can be
rebuilt from authoritative state. The executable fixtures prove those native
transitions without claiming archival custody. A persisted Workflow Engine
handoff is now reloaded through the Tasks aggregation surface and remains
visible until the authoritative Workflow transition completes. The journey
still needs pinned-composition reconstruction evidence and one target-tested
archive profile.
## Why this sequence
The sequence grows one connected product rather than advancing repositories in
isolation:
```text
Campaign reference
-> function-bound Postbox delivery
-> data-backed templates and reports
-> governed analytical data products
-> collaborative document lifecycle
```
Each stage reuses the preceding contracts:
- Campaign proves governed recipient selection, files, profiles, preview,
durable delivery, retry, reconciliation, audit, and role-specific UI.
- Postbox generalizes delivery from an email address to an institutional
responsibility that survives personnel changes.
- Templates and Reporting turn governed source data into reproducible outputs
and provide a safe launch boundary for specialist systems such as HIS.
- The BI path scales those source, transformation, quality, lineage, and report
contracts from one document to reusable analytical data products.
- DMS collaboration turns generated or uploaded artifacts into governed living
documents without confusing storage, editing, approval, or records
ownership.
## Continuous foundation lane
This lane is not a sixth product stage. It advances only as far as the next
journey needs and supplies contracts shared by all five stages.
1. **Governed integrations.** Connector profiles declare endpoint, authority,
source-of-truth mode, capabilities, health, limits, credential references,
and lifecycle. Network destinations are resolved and pinned at connection
time; private-network access is deployment-controlled and unpinned private
transports fail closed. Deleting a connector or provider immediately
deletes every secret it owns and records a non-secret audit event in the
same lifecycle action; an unavailable external secret provider blocks the
deletion or retirement rather than leaving an unaudited orphan.
2. **External effects.** Requested action, durable command, observed effect,
unknown outcome, retry, reconciliation, compensation/correction, and manual
intervention remain distinct and auditable.
3. **Institutional identity.** Identity owns subjects, Organizations owns units
and functions, IDM owns identity-to-function assignments and upstream
synchronization, and Access owns permission and acting-context decisions.
4. **Stable composition.** Modules cooperate through versioned Core-mediated
capabilities and typed references. Optional sibling modules are never
imported as implementation dependencies.
5. **Evidence and provenance.** Every imported datum, generated artifact,
delivery, transformation, and later correction can identify its source,
version, actor or system trigger, policy, and relevant execution evidence.
6. **Focused experience.** Central components and the interface pattern
language are mandatory. Views present only what the current task and actor
need while preserving an explainable escape to the configured full system.
7. **Adaptive documentation.** The owning module contributes canonical subject
matter; GovOPlaN Docs composes it for the installed modules, enabled
configuration, role, route, and task.
8. **Release proof.** Version alignment, clean install/upgrade, module
permutations, backup/restore, degraded-provider behavior, security checks,
and target-environment acceptance gate product claims.
9. **Artifact and infrastructure egress.** Tenant connectors use the pinned
connector boundary. Installer artifacts use a separate deployment-owned
trust path: approved origins, bounded pinned download or operator prefetch,
signature/checksum verification, safe extraction, and offline package-manager
execution. Database, broker, cache, and worker channels are constrained by
deployment network policy and authenticated transport rather than treated as
tenant connector profiles.
10. **Information governance.** Temporal browsing, purpose-aware access,
retention/legal-hold behavior, and institutional acting context are applied
to every owned object type. Historical reads use current authorization.
Module manifests state `contract_only`, `partial`, `enforced`, or
`not_applicable` adoption with evidence; supported maturity is blocked until
every applicable dimension is enforced.
11. **Durable human work.** Tasks aggregates explicit work and module-owned
attention items; Workflow Engine persists process state and handoffs;
Notifications attracts attention; Views focuses the relevant surfaces.
Leaving or refreshing the browser never becomes the only record that work
remains unfinished.
12. **Task-local tools.** Mail, Postbox, Templates, Files, and other common
actions may contribute bounded launch surfaces with return context. The
shell and Workflow compose them without copying their data or validation.
The optional Quick Access module presents configurable Work, Calendar,
Messages and Files categories; system/tenant/user settings and View/Policy
ceilings resolve their availability and ordering.
## Documentation contract for every reference stage
Documentation is one connected knowledge model rendered through different
perspectives, not separately maintained manuals that drift.
Every demonstrated journey provides:
- **Task guidance:** “I want to …” instructions using visible UI wording.
- **Process view:** actors, inputs, states, decisions, handoffs, exceptions,
completion, correction, and retention.
- **Concept view:** domain language and ownership boundaries.
- **Administrator view:** permissions, policies, profiles, configuration,
module requirements, migrations, and audit effects.
- **Operator view:** health, queues, retries, reconciliation, backup/restore,
incident handling, and target-provider checks.
- **Integration view:** capabilities, DTOs, APIs, events, idempotency,
compatibility, deep-link/launch contracts, and example payloads.
- **Security, privacy, and audit view:** threats, disclosure rules, secrets,
provenance, evidence, retention, and destructive actions.
- **Acceptance view:** runnable examples, expected results, failure injection,
and release gates.
- **Channel and records view:** assisted/non-digital intake and output,
representation, provenance, filing, retention, legal hold, and archive
consequences where the journey creates evidence or a record.
The Docs module selects and links these views according to installed
capabilities and actor context. Feature repositories remain the source of
truth for their behavior; Docs owns indexing, conditions, safe disclosure, and
presentation. Generated API/contract reference may supplement, but never
replace, the maintained behavioral explanation.
## Stage 1: Campaign as the demonstration module
### Product promise
> As a communications team, we can compose a campaign from governed recipient
> data, managed files, reusable mail configuration, and policy; preview the
> exact effect; authorize delivery; and resolve every recipient to a known or
> explicitly uncertain outcome without exposing transport internals to normal
> readers.
Campaign is the first reference composition, not an all-purpose mail client or
contact database.
### Ownership and composition
- Campaign owns campaign intent, content, campaign-local recipient snapshots,
personalization, execution state, and delivery evidence.
- Mail owns reusable profiles, credentials, protocol adapters, profile policy,
send/append execution, and provider-facing diagnostics. Campaign persists
only a stable Mail profile reference, never SMTP/IMAP settings or secrets.
- Files owns stored objects, provider connectors, file policy, and provenance.
Campaign persists stable attachment references and frozen execution evidence.
- Addresses or another directory may supply recipients, but Campaign freezes
the execution-time recipient and source evidence needed to explain the send.
- Policy, Access, Audit, Docs, Notifications, Organizations, Identity, IDM, and
Postbox contribute only through declared capabilities when installed.
### Implementation slices
1. Enforce the Mail-profile-only boundary and reject inline transport material
on authoring, import, update, build, and delivery paths. Legacy records are
preserved but fail closed until explicitly migrated.
2. Finish the central-component UI pass and consistent task-focused author,
reviewer, operator, and reader surfaces. Normal readers receive business
state, not paths, backend keys, worker claims, or raw provider diagnostics.
3. Prove test, single-send, resend, batch, append-to-Sent, retry,
reconciliation, quarantine, and manual intervention semantics against a
target SMTP/IMAP environment.
4. Complete adaptive Campaign, Mail, and Files documentation under the shared
documentation contract, including runnable reference campaigns and failure
drills.
5. Package the composition with sample data, policies, roles, configured views,
preflight, health checks, and pinned compatible versions.
### Gate
A clean installation can run the maintained examples through target providers;
no Campaign payload contains transport credentials; worker/provider failures do
not cause blind duplicate delivery; each supported role sees an appropriate
surface and explanation; install, upgrade, restore, and release evidence are
repeatable.
## Stage 2: Postbox and delivery to institutional responsibility
### Product promise
> As a sender, I can deliver a message to a named function in an organizational
> unit without knowing which individual currently performs it. As the current
> function holder or authorized delegate, I can discover, read, and act on that
> message; when assignments change, access follows current responsibility while
> the delivery and access history remains explainable.
### Ownership and composition
- Postbox owns addressable in-platform postboxes, messages, bindings,
participants, attachment references, access-sensitive events, and retention
inputs.
- Organizations defines units and functions. Identity defines subjects. IDM
assigns identities to functions and reconciles upstream directories. Access
resolves roles, delegations, acting context, and permissions.
- Mail is an optional external channel bridge. A Postbox is not an SMTP/IMAP
account and is never owned by one login credential.
- Campaign consumes a generic delivery-target capability. It may target an
email recipient or a function-bound Postbox without importing either
module's internals.
- A Postbox exists independently of its current holders. Vacancy, one
incumbent, several incumbents, and time-bounded delegation are valid states;
none turns the Postbox into a personal account.
- The E2EE target uses per-content keys and function/postbox key epochs.
Assignment grants device-bound access to policy-selected history, while
hand-over, revocation, and delegation expiry rotate or withdraw future key
access without claiming to erase plaintext already obtained.
- Content and signed manifests are immutable. Correction or replacement is a
linked new object/version, never silent substitution.
### Implementation slices
1. Stabilize Postbox manifest, permissions, DTOs, migrations, directory,
binding administration, access decisions, message create/list/read, and
audit events.
2. Define a typed delivery target and receipt contract for a Postbox id or an
organization-unit/function address. Resolution never copies current holders
into permanent mailbox ownership.
3. Complete the OrganizationsIdentityIDMAccess function-assignment and
delegation path, including provenance and compatibility migration from
legacy projections.
4. Define the E2EE key-grant/epoch profile for vacancy, multiple incumbents,
hand-over, history access, time-bounded delegation, recovery, and audit.
5. Add Campaign delivery to a function-bound Postbox, file attachment
references, optional external-mail bridge, and notification attention
signals through capabilities.
6. Document reassignment, delegation, vacancy, expired assignment, unavailable
optional module, retention, and future E2EE limitations honestly.
### Gate
A campaign message can be delivered once to a function-bound Postbox; an
authorized holder can act in an explicit context; a reassignment changes future
access without moving or rewriting the message. A vacant Postbox remains
addressable, retains the delivery, and visibly reports that no holder can
decrypt or act until assignment; multiple incumbents and delegates receive
independent key/access evidence. Delivery, key access, and later corrections
remain auditable.
## Stage 3: Templates, reports, data sources, and deep launches
### Product promise
> As a staff member arriving from a specialist system such as HIS, I can open a
> GovOPlaN report or document task with a governed source context already
> selected, review the effective data and template version, generate the
> document, and return or link the immutable result without re-entering data.
### Ownership and composition
- Templates owns reusable definitions, versions, merge schemas, localization,
render profiles, preview, and rendering.
- Reporting owns report definitions, parameters, curated data selection,
execution history, dashboards, sharing, scheduling, publication, and export.
- Connectors own protocol-specific access. A governed data-source catalog owns
connection profiles, source contracts, freshness, and health if repeated
use proves that a separate module is warranted.
- Files stores generated bytes. DMS owns a generated artifact once it becomes a
governed living document. Records owns later retention/archive semantics.
### Safe launch and generation contract
1. The source program opens a stable GovOPlaN route with an opaque, short-lived,
single-purpose launch reference—not credentials, arbitrary SQL, or trusted
personal data in the URL.
2. GovOPlaN authenticates the actor, authorizes the requested report/template
and source object, resolves the launch context server-side, and shows source,
freshness, purpose, and any blockers.
3. A curated connector/read-model capability returns a versioned input snapshot
that satisfies the template or report schema.
4. Generation records template/report definition version, parameters, input
snapshot or reproducible source references, transformation version, output
checksum, actor, and policy.
5. The result is stored or handed back through an explicit callback/reference
contract. Retries are idempotent; expired or replayed launch references fail
safely.
### Implementation slices
1. Implement the first Templates manifest, version/schema DTOs, safe preview,
deterministic render, package fragments, and Files-backed output.
2. Implement the first Reporting manifest, parameter/source DTOs, one
module-owned read model, execution evidence, and downloadable export.
3. Define the shared data-source/profile and launch-context contracts with one
read-only target connector and a mock HIS-style launch producer.
4. Deliver one end-to-end reference document and one analytical report from the
same governed source contract.
5. Add role-specific UI/docs, source freshness and validation states, deep-link
integration examples, and target-system acceptance tests.
### Gate
A user can follow a signed or server-side launch reference from a target system
to a prefilled GovOPlaN task, verify the source and freshness, render an output,
and reproduce why that exact artifact was produced. The browser cannot inject
arbitrary queries or transport secrets, and authorization is re-evaluated in
GovOPlaN.
## Stage 4: Governed BI and reporting
### Product promise
> As an institutional analyst, I can combine approved operational sources into
> versioned, quality-checked analytical data products, define transparent
> measures and reports, drill from aggregates where policy permits, and explain
> the source, transformation, effective organizational hierarchy, and reporting
> date behind every result.
### Lessons taken from SuperX
SuperX is useful inspiration because it combines many university sources in a
modular data warehouse, loads and transforms them on a schedule, keeps
historical or reporting-date views, supplies prepared subject modules, and
serves different aggregation levels through a common reporting surface. Its
official overview also emphasizes a lower operational-data layer, validation
logs, configurable load routines, organizational hierarchies, and transparent
report calculations. See the
[SuperX project overview](https://www.superx-projekt.de/__index.htm),
[administration and module architecture](https://www.superx-projekt.de/doku/kern_modul/admin/__index.htm),
and [cross-institution indicator module](https://www.superx-projekt.de/doku/kenn_modul/benutzer/__index.htm).
GovOPlaN should adopt the principles, not clone the implementation. Its distinct
value is connecting analytical results to governed work, responsibility,
evidence, policy, and correction across independently installable modules.
### Architecture layers
1. **Source catalog:** governed profiles, owners, purpose, schema/version,
classification, credentials, health, extraction mode, and source-of-truth
declaration.
2. **Ingestion and staging:** immutable run evidence, source snapshots or
watermarks, schema-drift detection, quarantine, replay, and personal-data
minimization.
3. **Transformation and quality:** versioned transformations, dependencies,
tests, validation findings, correction policy, and lineage. Silent repair of
source facts is not acceptable.
4. **Semantic products:** documented dimensions, measures, official key
mappings, organizational hierarchies, time/stichtag semantics, ownership,
access policy, and version.
5. **Consumption:** Reporting definitions, dashboards, drill-down, scheduled
outputs, APIs, open-data handoffs, and template generation.
6. **Operations:** scheduling, backfill, freshness objectives, observability,
cost/resource limits, retention, backup/restore, promotion between systems,
and reproducible release recipes.
### Selected first subject-area direction
- Start with read-only HIS and CampusOnline source profiles and produce the
institution's internal student-statistics data product plus reports derived
from it.
- Model reference/effective date separately from extraction, validation, and
freeze dates. For example, a semester state may describe `1 September`, be
accepted and frozen on `12 December`, and remain the reproducible source for
that semester's official statistics. Later corrections create a superseding
version; they never rewrite the frozen state silently.
- Make a graphical, typed data-flow editor the primary analyst surface. Nodes
represent governed sources/snapshots, validation, mapping, transformation,
aggregation, a list report/data product, template rendering, and publication.
The graph is versioned, testable, previewable, acyclic, and shows schema,
lineage, quality findings, policy, and pinned upstream versions.
- A list report may become a governed source for another flow. This enables
`source + transformations/aggregation -> list report/data product -> template
-> publishable report` without copying the calculation or losing lineage.
### Implementation slices
1. Bound the first HIS/CampusOnline student-statistics dataset and internal
report; define its legal/policy basis, source owners, official keys,
reference/freeze semantics, privacy level, and acceptance calculation.
2. Generalize the Stage 3 source profile into catalog, ingestion-run, staged
dataset, validation, transformation, lineage, and data-product contracts.
3. Implement one scheduled load and explicit frozen-state lifecycle with
quarantine/replay and a fully transparent transformation into the internal
student-statistics product.
4. Implement the first graphical typed flow editor over those proven nodes,
including version/diff, validation, preview, cycle prevention, and pinned
upstream product versions.
5. Serve the product through Reporting with policy-aware aggregate and
drill-down behavior; export reproducible evidence and calculation metadata.
6. Chain one list report as a source into a template-backed publishable report.
7. Package mappings, transformations, validations, reports, documentation, and
tests so they can be promoted from development to test to production.
### Gate
One internal student-statistics product can be rebuilt from declared HIS and/or
CampusOnline snapshots, preserves its reference and freeze dates, passes
explicit quality tests, explains every measure and mapping, enforces disclosure
policy, survives schema/freshness failures visibly, and produces the same
accepted list and template-backed report across promoted environments.
## Stage 5: Collaborative document editing
### Product promise
> As an authorized group, we can turn an uploaded or generated artifact into a
> shared document, edit it concurrently or through controlled check-out,
> comment, review, approve, compare, and recover versions, then freeze the
> accepted rendition as evidence or a record without losing who changed what.
### Ownership and composition
- Files owns bytes, checksums, storage providers, upload/download, and low-level
file permissions.
- DMS owns document identity, versions, renditions, editing sessions, locks,
comments, reviews, approvals, signatures, external DMS references, and
document-level evidence.
- A collaboration connector owns provider-specific Collabora, OnlyOffice,
Nextcloud/OpenDesk, or another office-suite protocol. DMS owns the session
lifecycle and authorization even when editing occurs externally.
- Templates may create the first version. Cases, Campaign, Postbox, Reporting,
and Workflow may reference documents. Records owns classification, legal
hold, archive handoff, and disposal after the document enters that lifecycle.
### Implementation slices
1. Implement DMS document/version/reference DTOs and one Files-backed version
path, including optimistic concurrency, checksums, locks, and audit.
2. Add comments, review requests, approval state, immutable accepted
renditions, comparison, and recovery of prior versions.
3. Define a provider-neutral editing-session capability with short-lived
grants, callback authentication, save idempotency, health, and reconciliation.
4. Integrate one external collaborative editor, including loss of provider,
concurrent save, stale callback, permission revocation, and restore tests.
5. Connect generated reports/templates, Postbox/Campaign sharing, focused
document views, configured docs, and Records handoff.
### Gate
Multiple authorized actors can edit a document without silent lost updates;
every accepted version has stable content and provenance; provider callbacks
cannot bypass current authorization; degraded or ambiguous saves are visible
and reconcilable; the approved rendition can enter records without making Files
or the external editor the document-lifecycle owner.
## Program-wide acceptance and pause rules
- Do not start a broad generic platform abstraction until two concrete stages
demonstrate the repeated contract.
- Do not make Workflow a dependency of these stages. Manual transitions,
explicit actions, and focused views should become stable providers that a
later transition engine can coordinate.
- Do not treat a remote system as trusted merely because it supplied a deep
link, callback, webhook, file, identity, or data row.
- Do not claim a stage complete from local unit tests. Use pinned composition,
target integration, failure drills, adaptive docs, and operator evidence.
- Do not claim a module complete while its relevant information-governance
dimensions remain `contract_only` or while the reference journey lacks an
assisted-channel and records outcome where those are applicable.
- A later stage may prototype contracts while the preceding gate is being
proven, but it may not redefine an owning module's boundary by convenience.
## Decisions intentionally deferred
The sequence itself is selected. These bounded choices remain for the stage
that first needs them:
- first target SMTP/IMAP, file/directory, identity, and deployment profile;
- first external IDM source and identity conflict/disable policy;
- first trusted Postbox assurance/recovery/history-access profile and
external-mail bridge;
- first HIS or other launch producer and callback/reference contract;
- exact HIS/CampusOnline source endpoints, student-statistics keys and accepted
calculation, freeze/correction policy, privacy profile, and permitted
drill-down level;
- datasource provider selection and quality/promotion policy; the architecture
now separates `govoplan-datasources` lifecycle from `govoplan-connectors`
acquisition and `govoplan-dataflow` transformation;
- first collaborative editor/provider and whether the first UX is concurrent
editing, controlled check-out, or both; and
- first Records/archive target and approval/signature assurance level.
+100
View File
@@ -0,0 +1,100 @@
# GovOPlaN Roadmap
## Purpose
GovOPlaN should become the connective, governance-aware operating layer of an
institution: people complete services and work without learning the module
graph, while the institution can explain authority, policy, source data,
effects, evidence, and recovery.
This is the concise product roadmap. It states durable outcomes and sequence,
not release dates or issue state. Use [Strategy Status](STRATEGY_STATUS.md) for
the current reconciliation and Gitea issues for active work. The
[detailed connected-platform vision](reference/CONNECTED_GOVERNANCE_PLATFORM_ROADMAP.md)
retains stakeholder perspectives, configuration archetypes, and the complete
outcome-story catalogue.
## Product Promise
GovOPlaN will:
1. model institutional context, responsibility, authority, and time;
2. turn incoming information into owned, reviewable human and machine work;
3. connect native and external systems without obscuring the source of truth;
4. preserve decisions, effects, records, corrections, and recovery evidence;
5. support digital, assisted, paper, message, calendar, and system channels as
paths through the same governed work; and
6. package successful configurations so institutions can adopt them without
code forks or loss of local autonomy.
It will not replace every specialist system, copy all data into one master
database, infer authority from membership, or claim production maturity from
repository breadth.
## Outcome Horizons
| Horizon | Outcome | Completion evidence |
| --- | --- | --- |
| Trustworthy baseline | A pinned composition can be installed, upgraded, operated, explained, and recovered. | Signed artifacts, clean install/upgrade, provider failure tests, restore drill, coherent UI, and target evidence |
| Connected work | Intake becomes accountable work with context, assignment, review, communication, and evidence. | One digital and assisted service reaches a decision and eAkte without losing responsibility or state |
| Reusable products | Complete service, communication, and data outcomes ship as governed configuration packages. | Two materially different deployments adapt packages without code forks |
| Institutional assurance | Records, transparency, privacy, risk, regulated review, and reporting connect to real operations. | A consequential decision can be reconstructed, corrected, retained, and disclosed under policy |
| Federated ecosystem | Autonomous installations exchange signed data and configuration across explicit trust boundaries. | Paired-instance exchange, reconciliation, supported deployment profiles, and independent evidence |
## Current Sequence
The sequence is outcome-led. Shared foundation work enters when one of these
proofs needs it.
1. **Enforce the platform quality contract.** German is the reference locale;
help, accessibility, temporal browsing, purpose-aware access, retention,
institutional context, optional-module combinations, and recovery behavior
become measurable release gates.
2. **Complete governed communication.** Prove recipient selection, Campaign,
Files, Mail, function-bound Postbox delivery, acknowledgement, uncertain
outcomes, correction, filing, and recovery against a named target.
3. **Complete the monthly-data and sanctions journey.** Acquire immutable
source snapshots, validate and reconcile data interactively, preserve
lineage and review, publish reports and files, and deliver accepted results.
4. **Complete inclusive service to decision.** Accept digital or assisted
input, establish actor and purpose, persist human handoffs, decide, notify,
and reconstruct the exact eAkte under current authorization.
5. **Complete discovery and external coexistence.** Finish native PostgreSQL
search coverage, prove reauthorization and reindexing, then prove one
external product connector and one paired GovOPlaN federation exchange.
6. **Prove production operation.** Complete multi-host, provider, restore,
accessibility, volume, key-custody, and independently signed target
evidence before raising maturity claims.
## Continuous Foundation
Every journey applies the same boundaries:
- modules cooperate through versioned Core contracts and typed references;
- permissions, policy, institutional context, purpose, and current authority
are evaluated before presenting or acting on data;
- requested actions, durable intent, observed effects, unknown outcomes,
retries, reconciliation, and correction remain distinct;
- Workflow Engine coordinates stable module-owned actions and human handoffs;
it does not become a second owner of domain state;
- Files owns managed bytes, Records owns institutional filing and retention,
and source systems retain explicitly declared authority;
- focused views and product areas reduce interface complexity without granting
access or hiding material consequences;
- configuration packages include terminology, forms, policies, workflows,
views, reports, providers, documentation, migration, and evidence; and
- maturity advances from scaffold to vertical slice, reference-ready,
supported, and LTS only with evidence appropriate to each claim.
## Decision Rule
A roadmap item should answer all of the following before implementation:
1. Which real journey and actor outcome does it improve?
2. Which module or external system owns each object and source of truth?
3. Which institutional, temporal, purpose, and policy context applies?
4. Which effects, evidence, retention, failure, and recovery states result?
5. Which package and target evidence will prove the outcome?
If those answers are missing, retain the idea in the Product Input Register or
Gitea discovery work rather than opening an unbounded implementation program.
+132
View File
@@ -0,0 +1,132 @@
# GovOPlaN Strategy Status
## Status Record
| Field | Value |
| --- | --- |
| Reconciled on | 2026-08-06 |
| Source scope | Local workspace manifests, source inventory, focused journey checks, signed release evidence, and live Gitea issue state |
| Stable direction | [Platform Core Ideas](PLATFORM_CORE_IDEAS.md) and [Roadmap](ROADMAP.md) |
| Collected product input | [Product Input Register](PRODUCT_INPUT_REGISTER.md) |
| Delivery source | Gitea issues |
This is the only prose source for current cross-product status. It is a
reconciliation, not a release certification. Module manifests and target
evidence remain authoritative for specific maturity claims.
## Portfolio Snapshot
- 66 source module manifests were loadable and architecture-declared.
- 48 modules declared `vertical_slice`; 18 declared `scaffold`.
- No module declared `reference_ready`, `supported`, or `lts`.
- The live portfolio had 133 open issues, including 37 priority-P1 items.
- 118 open issues had no milestone, so issue labels do not yet express a
reliable completion sequence on their own.
- Three product package manifests existed: governed communication, governed
data and assurance, and service to decision. None had crossed the complete
target-evidence gate.
These counts are dated. Refresh them rather than copying them into another
document.
## Interface And Contract Evidence
The 2026-08-06 source inventory found:
- 1,307 UI fields and 1,291 UI actions;
- 8,157 stable interface declarations with no duplicate IDs;
- 43 frontend routes and 920 backend endpoints;
- no public WebUI surfaces missing runtime declarations;
- no stale runtime route declarations;
- no unclassified endpoint without a static UI reference;
- all 1,307 fields with a resolvable F1 context; 1,143 remain candidates for
richer field-specific content beyond page/module fallback;
- German (`de`) as the complete reference locale and no used key missing from
the required German or English catalogs;
- 3 module information-governance dimensions classified as `enforced`, 1 as
`partial`, and 260 as `contract_only`.
This is an honest platform-wide baseline, not a claim that temporal,
purpose, retention, and institutional-context adoption is complete.
## Credible Current Outcomes
### Platform foundation
Module discovery, optional dependency validation, migrations, shared WebUI,
tenant and access foundations, signed catalogs/packages, event delivery,
recovery contracts, contextual help, views, temporal titlebar context, and
stateless-runtime patterns are implemented and tested at varying depths.
### Governed communication
Campaign authoring, recipient data, attachments, templates, mail profiles,
mock/real delivery paths, audit evidence, reporting, distribution-list
composition, and optional Postbox delivery form the deepest product cluster.
Target provider, accessibility, recovery, and high-volume evidence still
prevent a reference-ready claim.
### Institutional service and decision
Services, Forms, Forms Runtime, Cases, Parties, Mandates, Approvals, Committee,
Voting, Decisions, Portal, Postbox, and Audit have an executable service-to-
decision fixture. Public and invitation intake can retain Files-backed
evidence; Forms submissions, Cases, and formal Decisions can be explicitly
filed as exact eAkte source revisions and reconstructed through permission-
rechecked native Search projections. A durable Workflow Engine handoff now
survives session restart and appears through the Tasks work inbox until the
authoritative transition completes. Browser-complete assisted intake, broader
work projections and escalation, production identity and delivery, a named
archive profile, and target evidence remain.
### Governed data and assurance
Connectors, Datasources, Dataflow, Reporting, Search, Policy, Risk Compliance,
and Workflow provide source governance, immutable snapshots, transformation,
quality, semantic reporting, and provenance foundations. The monthly-data and
sanctions compositions now prove immutable connector snapshots, pinned
Dataflow publication, Risk Compliance review, and rescreening in process. The
journeys still need target connector profiles, complete interactive
reconciliation, governed export/delivery, and browser-level handoff evidence.
## Material Gaps
| Gap | Consequence | Next proof |
| --- | --- | --- |
| No reference-ready product package | The platform cannot yet make a bounded supported-product claim | Complete one named target composition and evidence bundle |
| Human-work spine is only an MVP | Tasks aggregates explicit work plus Workflow, Approval, and unread Postbox projections, but broad domain coverage, deadline escalation, assignment lifecycle, and focused product UX remain | Extend source providers through the three reference journeys and prove overdue/reassignment behavior in browser tests |
| Records/eAkte target integration incomplete | Native lifecycle, retention, holds, approval, recovery, and transfer simulation are implemented, but real custody is not proved | Target-test one archive/xdomea profile and complete the assisted reference journey |
| Cross-cutting governance adoption uneven | Historical and purpose-sensitive behavior varies by module | Enforced adoption declarations and route/query/effect migration |
| Explicit help/accessibility depth incomplete | German/reference and F1 association gates now pass, but generic fallback remains too common | High-risk German help content and browser/a11y matrix |
| Real federation absent | Cross-institution exchange remains connector-specific | Paired-instance signed exchange and reconciliation proof |
| External production evidence incomplete | Scale, restore, interoperability and custody claims remain conditional | Real target drills and independent signed evidence |
## Active Strategic Order
1. Establish German, help, temporal, purpose, retention, and institutional
context as enforceable platform quality contracts.
2. Complete governed communication and Postbox against a named target.
3. Complete the monthly-data flow and use it as the data foundation for
sanctions screening.
4. Complete the browser and resumable-work proof for the digital and assisted
service-to-decision journey with its existing exact eAkte filing contracts.
5. Complete native PostgreSQL search coverage for remaining journey-owned
objects and prove reauthorization and reindex operations at target volume;
keep OpenSearch optional. Communication, Records, service-to-decision,
Dataflow, Reporting, Risk Compliance, and Datasource catalogue sources now
exist.
6. Prove one external product connector and one GovOPlaN federation exchange.
7. Finish multi-host, restore, provider, accessibility, and independent signed
target evidence before increasing maturity claims.
## Refresh Procedure
Refresh this page only from evidence:
1. run `tools/checks/check-manifest-shapes.py`;
2. run `tools/inventory/platform-interface-inventory.py --strict
--strict-declarations --strict-endpoints`;
3. run the selected reference-journey checks;
4. inspect signed release and target evidence;
5. query live Gitea issue/milestone state;
6. update the dated values and material gaps here;
7. retain prior assessments as dated evidence rather than rewriting them.
@@ -0,0 +1,212 @@
# System Administrator Lifecycle User Story
## Story
> As a system administrator, I can execute one shell command that downloads a
> verified GovOPlaN distribution and starts a completely configured Core control
> plane with the official package directory available but only the protected
> baseline active. In the WebUI I can browse compatible signed
> module releases, select the modules for this installation, and follow every
> download, validation, migration, installation, activation, and health-check
> step. When an update is available, I can review its impact and confirm it.
>
> I can add API or worker replicas on containers or other servers without
> copying mutable local state. I can operate development, test, and production
> systems, transfer a versioned configuration through the WebUI, undo an applied
> configuration version, test an update in one environment, and then apply the
> same immutable update recipe to another environment.
This is a product-level story owned by the GovOPlaN platform rather than by an
individual domain module. It joins installation, module lifecycle, operations,
configuration packages, and release provenance into one administrator journey.
The canonical backlog item is
[GovOPlaN #13](https://git.add-ideas.de/GovOPlaN/govoplan/issues/13).
## Terms
- **Core control plane:** the smallest bootable distribution: Core API, Core
WebUI, PostgreSQL, Redis, installer worker, migration runner, and durable
storage configuration. An immutable image may carry the full verified package
profile, but optional modules are not active or tenant-entitled by implication.
- **Bootstrap administrator:** a single-use, time-limited installation identity
that may access only first-run and module-lifecycle functions. It is retired
when the selected identity/access configuration becomes healthy.
- **Module release:** an immutable, signed package plus manifest, compatibility
contract, migrations, WebUI contribution, checksums, and SBOM references.
- **Configuration revision:** an immutable, schema-versioned export of
non-secret system/module settings, policies, compositions, and secret
references. Secret values are never exported.
- **Update recipe:** the reviewed, immutable plan containing exact Core/module
versions, configuration revision, migration order, preflight results,
maintenance/drain requirements, health checks, and permitted rollback or
forward-recovery actions.
## Acceptance journeys
### One-command first installation
1. The administrator runs one documented command on a supported host.
2. The bootstrapper verifies a signed distribution manifest before executing or
starting downloaded artifacts.
3. It checks container/runtime, ports, storage, entropy, memory, architecture,
and connectivity requirements; generates deployment-local secrets with
restrictive permissions; and never prints them.
4. It starts PostgreSQL, Redis, Core API/WebUI, one installer worker, and the
migration runner. Readiness does not pass until migrations and durable
dependencies are healthy.
5. It prints the local URL and one-time bootstrap credential. Re-running the
command is idempotent and shows or repairs the existing installation rather
than creating another identity or database.
6. Only the protected baseline is enabled. Installed package availability does
not grant permissions, tenant entitlement, View visibility, or capability
opt-in.
### Module selection, installation, and update
1. Core reads an approved signed catalog and trusted keyring through the shared
outbound-network policy.
2. The WebUI shows available, installed, compatible, blocked, withdrawn, and
update-available releases with their channel, provenance, contracts,
migrations, permissions, configuration requirements, and release notes.
3. Selecting modules produces a dependency/compatibility plan before any
mutation. The administrator can amend the selection or confirm the plan.
4. Installation executes durably in a worker. The UI receives persisted step
state and can reconnect without losing progress.
5. Package signature/checksum, version alignment, contract closure, migration
graph, configuration schema, and health checks are mandatory gates.
6. Remote artifacts are acquired by a deployment-owned downloader that permits
approved origins, pins the validated connection peer, enforces size and time
limits, and verifies signature/checksum before making a local immutable
artifact available. `pip`, `npm`, archive tools, and any compatibility Git
importer then run without unrestricted network access. A restricted-network
profile can satisfy the same contract through operator-prefetched artifacts.
7. A failed or interrupted operation reaches a visible retry, forward-recovery,
rollback, or manual-intervention state. It never reports success merely
because the initiating request returned.
8. An available update follows the same preview and confirmation path. Exact
repeated requests are idempotent.
### Horizontal scaling
1. API, WebUI, installer, scheduler, and ordinary worker roles are stateless
with respect to local container disks. Durable state uses PostgreSQL, Redis,
and configured shared file/object storage.
2. A documented command can add API or queue-specific worker replicas. The
default Compose profile supports local scale-out; an orchestrator profile
supplies equivalent health/readiness probes and rolling replacement.
3. Only one migration or module-lifecycle mutation may own the deployment lock,
while any healthy replica can serve read and normal domain traffic.
4. Workers announce identity, queues, software composition, heartbeat, and
drain state. Operators can see skew and safely retire a replica.
5. Sessions, throttling, idempotency, jobs, installer progress, and scheduled
work remain correct when requests move between replicas.
### Development, test, and production promotion
1. The administrator exports a configuration revision from development through
the WebUI. The package is versioned, checksummed, attributable, and contains
secret references or required-secret declarations, never secret values.
2. Test imports the package into a preview area. Core reports environment-bound
values, missing capabilities/secrets, compatibility changes, and the exact
apply plan.
3. Applying creates a new revision; the previous effective revision remains
addressable. Undo is a new audited revision that restores the earlier
configuration where contracts permit it.
4. After tests and health checks pass, Core emits an update recipe with exact
immutable release and configuration identifiers.
5. Production validates the recipe against its own environment, requires a new
confirmation, and executes the same ordered plan. Environment-specific
secret bindings and endpoints remain local.
6. Promotion never copies tenant/business data implicitly. Data migration,
anonymized fixtures, and backup/restore are separate explicit operations.
## Safety and governance requirements
- Catalogs, artifacts, recipes, and configuration revisions have signatures,
checksums, provenance, expiry/revocation semantics, and audit evidence.
- Catalog and artifact retrieval is not delegated to an unrestricted package
manager. Approved-origin and peer-pinned download, response bounds,
signature/checksum verification, extraction safety, offline installation, and
deployment egress policy form separate defenses.
- The installer has a dedicated narrow authority; normal domain permissions do
not imply host/package-management access.
- The UI distinguishes reversible configuration rollback from database or
package migrations that require forward recovery.
- Cluster-wide mutations use fencing/leases so an expired worker cannot later
commit a stale result.
- Update plans declare availability impact, worker draining, backup/restore
prerequisites, database compatibility windows, and post-change probes.
- Deployment profiles set secure headers, trusted proxies/hosts, body limits,
outbound-network policy, storage, TLS/cookie posture, and observability.
- Release publication requires aligned backend/frontend/manifest versions and
machine-readable dependency/SBOM provenance.
## Implementation slices
Implementation status as of the current source tree:
- Slice 1 has a published production-artifact baseline. Immutable
[`v0.1.14`](https://git.add-ideas.de/GovOPlaN/govoplan/releases/tag/v0.1.14)
binds source commit `1f039dd39c1ce2672f4978c8abc6dff862ef1445`, a signed
one-file deployer, exact API/Web and managed-dependency image digests,
composition, SBOMs, and provenance. Runtime Distribution
[run #459](https://git.add-ideas.de/GovOPlaN/govoplan/actions/runs/459)
passed migrations, schema checks, non-root API/Web readiness, and worker
delivery/shutdown on both amd64 and arm64. Each future release must renew the
evidence, and a real installation must still produce topology-specific
ingress, failover, backup, and recovery receipts.
- Slice 6 has a working application-tier foundation: state profiles, shared
object storage, runtime node registration/heartbeats/drain, fenced scheduler,
migration serialization, exact-head startup waiting, Ops visibility, and a
Kubernetes export. Production acceptance still requires topology-specific
failover and restore drills.
- The recovery foundation for slices 4 and 5 is implemented as a Core recovery
ledger and deployment operation journal. Automatic database backup and broad
adoption by module-owned external effects remain open work.
1. **Reproducible Core-baseline distribution.** Publish pinned multi-architecture
full-package images, signed distribution manifest, Core-baseline Compose profile, bootstrap
preflight, generated secrets, readiness, and idempotent rerun/repair.
2. **First-run control plane.** Add the restricted bootstrap administrator,
one-time enrollment, initial catalog/keyring configuration, and retirement
after durable administrator access is established.
3. **Read-only online module directory (implemented foundation).** Admin falls
back to the signed public stable directory, presents installed/update state,
searchable availability/blocker filters, immutable source/artifact
provenance, configuration requirements, release notes, and technical
compatibility. Withdrawn releases remain visible but cannot be planned.
Operator-configured catalogs remain an explicit override.
4. **Durable module plan and install (implemented local boundary).** Catalog
selection creates a reviewed plan; the installer queue, lock, preflight,
maintenance gate, digest-verified artifact cache, rollback drill, and run
evidence remain separate from the API process. Shared deployments convert
the same intent into a new immutable release composition instead of mutating
one replica.
5. **Safe module update.** Add drain/maintenance coordination, backup gate,
migration compatibility window, reconnectable progress, health verification,
retry/recovery, and update notification.
6. **Stateless replica profile.** Continue module adoption and operational
proof for the implemented role commands, shared-state validation, runtime
registration/drain, fenced scheduler, and Kubernetes application-tier
export. Prove multiple API and worker replicas against the target shared
dependencies.
7. **Configuration revision model.** Define provider export/import schemas,
canonical serialization, secret references, validation/diff, immutable
revision storage, audit, apply, and undo-as-new-revision.
8. **Environment promotion and recipes.** Add source/target fingerprints,
preview, environment bindings, acceptance evidence, exact recipe generation,
signed transfer, and independently confirmed application.
9. **Operational proof.** Exercise interrupted installs, stale locks, unavailable
catalogs, revoked keys, failed migrations, replica loss, configuration undo,
and development-to-test-to-production promotion in release CI and target
drills.
## Explicit non-goals for the first distribution slice
- Activating, tenant-entitling, or exposing optional modules merely because the
immutable image carries their verified packages.
- Exporting secrets or production business data with configuration.
- Pretending every schema migration can be reversed automatically.
- Building a proprietary orchestrator instead of supporting Compose and a
standard cluster scheduler through the same role/readiness contracts.
- Allowing the browser process to execute arbitrary shell commands.
File diff suppressed because it is too large Load Diff