feat: promote parallel autonomous governance preset v0.2.0

docs/project-statistics.md

@@ -96,6 +96,7 @@
 | 2026-07-16 | `autonomous-run-governance` v0.2.2 Bedien- und Zustandsvertrag | 64 | — | 472 | Den Feature-029-Feldnachweis in ein ausfuehrliches, bilinguales CEFR-B2-Bedienhandbuch fuer Lernende und Maintainer ueberfuehrt: Zweck, Zielgruppen, Liefermodi, Konvergenz, Installation, vollstaendige Prompt-Beispiele, Status, Stop, Resume, Run-State-/Exact-Head-Evidence, Retrospektive, Lernreihenfolge und Fehlersuche. Die Paketvalidierung entdeckte zusaetzlich reproduzierbar, dass die lesbare Skill-Ueberschrift `Deliver` in Feature 029 als ungueltiger Maschinenzustand gespeichert worden war. v0.2.2 haelt die Validatoren streng und schreibt fuer Remote-Closeout eindeutig `Publish`, `Review` oder `MergeAndSync` vor. Der Kandidat umfasst vor diesem Ledger-Eintrag `+1175/-161`, also 1014 Nettozeilen ueber kanonisches und publizierbares Preset, Agent-/Constitution-Paritaet, AR-029-Workitems und Paketledger. Spec Kit 0.12.11, Sieben-Preset-Dev-Installation, 18 Contributions, Resolve, Disable/Enable/Remove/Reinstall, fuenf eindeutige Codex-Skills, beide Exact-Head-Validatoren sowie gueltige und ungueltige Stage-Fixtures sind lokal gruen. Konservative Manualreferenz: `14,7` Tage beziehungsweise `114,6` Stunden bei 80 Zeilen/Tag; Thorsten-Solo: `11,8` Tage beziehungsweise `91,7` Stunden bei 100 Zeilen/Tag. Home-Baseline-PR #70, oeffentlicher Preset-PR #7, Release `v0.2.2` mit gepruefter Tag-ZIP und TuiVision-Adoption PR #85 sind abgeschlossen; Feature 030 wurde nicht gestartet. |
 | 2026-07-16 | `autonomous-run-governance` v0.2.2 Hard-Abort-/Resume-Feldtest | 64 | — | — | Einen isolierten Spec-Kit-0.12.11-Lauf mit voller Sieben-Preset-Matrix, `LocalImplementation` und ohne Git-Remote nach `HARD_ABORT_READY` ueber die Codex-Oberflaeche abgebrochen. Der Kindprozess schloss Attempt 1 danach erfolgreich ab, waehrend der persistierte Zustand bei `Active` 13/30 und die Tasks bei 15/30 standen. Read-only Status klassifizierte effektiv `Interrupted`, veraenderte keinen Hash, und der allgemeine autonome Command verweigerte die implizite Fortsetzung. Explizites Resume revalidierte lokale Autoritaet, pruefte Completed-Marker und bytegenaues Ergebnis, startete keinen zweiten Attempt und endete mit `Retrospective`, `Completed`, 30/30 sowie zwei gruenen State-Validatoren. Retrospektive `NoPromotion`: v0.2.2 deckt State-, Evidence-, Duplicate-Work- und Berechtigungsgrenzen bereits ab; nur eine lokale EOF-Formatkorrektur war noetig. Kein v0.2.3-, Release-, Adoptions- oder Community-Issue-Update erforderlich; Feature 030 bleibt bis zu einem separaten Auftrag ungestartet. |
 | 2026-07-17 | Verbindliche Spec-Kit-Standard-Siebenermatrix | 65 | — | 484 | `autonomous-run-governance` v0.2.2 als siebtes Standard-Preset in zentraler Matrix, Installern, Registry-Profil, Constitution, Agenten-Dateien, Templates, Secure-Development-Basis 3.1.0 und Lernreihen verankert. Das bisherige Profil `standard-six-governance-presets` bleibt als deprecated Kompatibilitaetsalias mit exakt identischer Preset-Matrix erhalten. Installation erteilt weiterhin weder autonome Ausfuehrungs- noch Remote-, Merge-, Bypass- oder Provider-Berechtigungen. Bash-/PowerShell-Syntax, Generatoren, Matrixparitaet, Manifest und isolierte Neuinstallation samt Idempotenz sind geprueft; der Bash-Installer erkennt zusätzlich normale Klone und verknüpfte Git-Worktrees über `git rev-parse --git-dir`. |
+| 2026-07-18 | `parallel-autonomous-run-governance` v0.1.0 und nativer Real-Agent-Smoke | 66 | — | — | Experimentelles achtes Preset mit vier Topologien, UUID-Vertraegen, isolierten Worktrees, maximal drei parallelen Workern, DAG-Pruefung, gehashten Handoffs, kooperativem Stop/Resume und expliziter Alternativauswahl umgesetzt. Der vom Repository-Eigentuemer bewusst autorisierte native macOS-Entwicklungs-Smoke fuehrte 13 echte Codex-/Spec-Kit-Autonomous-Worker ohne Remote-Schreibzugriff aus: 13/13 Exitcode 0, Replicated/Independent/Alternative jeweils beobachtete Parallelitaet 3, Pipeline `A -> {B,C} -> D` mit Parallelitaet 2 und vier validierten Handoffs. Kanonisches Preset, Publikationskopie und Feature-Artefakte umfassen vor dieser Ledger-Zeile 4.180 Zeilen; konservative Manualreferenz `52,3` Tage beziehungsweise `407,6` Stunden bei 80 Zeilen/Tag, Thorsten-Solo `41,8` Tage beziehungsweise `326,0` Stunden bei 100 Zeilen/Tag. Das Preset bleibt bis zum nativen Secure-CaseTracker-Feldtest experimentell und opt-in; Container-First bleibt ausserhalb dieses ausdruecklichen Entwicklungs-Overrides verbindlich. |
 
 ---
 

specs/007-parallel-autonomous-run-governance/plan.md

@@ -0,0 +1,57 @@
+# Implementation Plan: Parallel Autonomous Run Governance
+
+## Architecture
+
+The preset adds a campaign layer above autonomous single runs. A tracked
+campaign manifest declares portable intent. A local runner-profile file binds
+agent and provider commands without storing secrets. Runtime worktrees, logs,
+locks, and process results live outside tracked feature artifacts.
+
+The PowerShell 7 coordinator is the canonical structured implementation. The
+Bash entry point delegates argument-for-argument to that implementation, which
+keeps behavior identical on macOS, Linux, and Windows where PowerShell 7 is part
+of the required workspace toolchain.
+
+## Contracts
+
+- `parallel-campaign.json`: immutable campaign intent and worker graph.
+- `parallel-runner-profiles.json`: local executable/argument-array bindings.
+- `parallel-campaign-state.json`: resumable aggregate lifecycle.
+- `worker-result.json`: worker outcome, head, evidence, and optional PR.
+- Existing `autonomous-run-state.json`: authoritative worker-internal state.
+
+## Delivery
+
+Implement and validate in Level 0, publish v0.1.0 in a protected standalone
+repository, then opt in the Secure CaseTracker coordinator and six language
+repositories. Promote to v0.2.0 and the standard eight-preset matrix only after
+the Units 00-03 field campaign succeeds.
+
+## Native Development Override
+
+Repository owner Thorsten Hindermann explicitly authorized both the 13-worker
+smoke set and the Secure CaseTracker Units 00-03 field campaign to run natively
+on the development Mac on 2026-07-18. This is a deliberate development-phase
+override of the otherwise binding Secure Trader Container-First rule.
+
+The campaign evidence MUST record the override, host platform, agent family,
+maximum configured and observed concurrency, repositories, run IDs, and start
+and completion times. Concurrency remains capped at three. The override ends
+after field findings are captured and does not apply to learner, production,
+routine maintenance, or later campaigns. Those return to Container-First
+unless the repository owner issues a new explicit authorization.
+
+## GitHub Billing Development Override
+
+Repository owner Thorsten Hindermann explicitly authorized the development
+field campaign to continue despite GitHub Actions jobs in the six private
+Secure CaseTracker repositories being rejected before their first step with
+the account-payment or spending-limit annotation.
+
+This exception is narrow: those exact startup rejections are recorded as
+external provider `N/A`, not as passing technical evidence. Every local build,
+test, security, accessibility, documentation, and autonomous-state gate remains
+required. Technical CI failures, actionable reviews, missing worker evidence,
+and any other unavailable check remain blocking. Ordered merges may use the
+owner's admin bypass only after the all-ready barrier and successful local
+gates. The exception expires with the native field campaign.

specs/007-parallel-autonomous-run-governance/spec.md

@@ -0,0 +1,84 @@
+# Feature Specification: Parallel Autonomous Run Governance
+
+**Feature Branch**: `007-parallel-autonomous-run-governance`
+**Created**: 2026-07-18
+**Status**: Approved for implementation
+
+## Goal
+
+Create an experimental Spec Kit preset that coordinates several existing
+`autonomous-run-governance` runs without weakening their lifecycle, evidence,
+or permission boundaries.
+
+## User Scenarios
+
+1. Run one language-neutral assignment in several isolated repositories.
+2. Run independent features concurrently without sharing branches or worktrees.
+3. Produce alternative solutions and require an explicit human selection before
+   consolidation.
+4. Execute a dependency-ordered agent pipeline with immutable, hashed handoffs.
+5. Stop, inspect, and resume a campaign without guessing whether child work
+   completed.
+
+## Functional Requirements
+
+- **FR-001**: The preset MUST use immutable UUID campaign and worker run IDs.
+- **FR-002**: Every worker MUST have an isolated branch and Git worktree.
+- **FR-003**: Concurrency MUST be limited to a declared value from 1 through 3.
+- **FR-004**: The campaign MUST support `ReplicatedTargets`,
+  `IndependentFeatures`, `AlternativeSolutions`, and `Pipeline`.
+- **FR-005**: Pipeline dependencies MUST form a directed acyclic graph.
+- **FR-006**: A worker MUST retain the existing autonomous-run state contract;
+  campaign state MUST reference rather than duplicate it.
+- **FR-007**: Runner commands MUST execute as executable plus argument array,
+  without shell evaluation.
+- **FR-008**: Ordinary worker failure MUST not cancel unrelated running work.
+- **FR-009**: Failed pipeline dependencies MUST block their descendants.
+- **FR-010**: Campaign-integrity, security, permission, or evidence failures
+  MUST stop new scheduling.
+- **FR-011**: Stop MUST be cooperative. Unexpected process loss MUST require
+  revalidation before retry.
+- **FR-012**: Alternative solutions MUST require a named human selection.
+- **FR-013**: Merge-and-sync delivery MUST enforce an all-ready barrier before
+  the first merge and MUST stop after the first merge failure.
+- **FR-014**: Handoffs MUST identify producer, consumer, path, and SHA-256.
+- **FR-015**: Installation MUST grant no execution or remote authority.
+- **FR-016**: Bash and PowerShell entry points MUST expose equivalent behavior.
+- **FR-017**: A sequential worker MAY inherit the validated exact head of one
+  direct dependency in the same repository without sharing its worktree.
+- **FR-018**: Reviewed campaign-specific operator instructions MUST be routed
+  visibly to every worker without weakening undeclared authority boundaries.
+
+## Success Criteria
+
+- **SC-001**: Deterministic fixtures pass for all four topologies.
+- **SC-002**: The scheduler never observes more than three active workers.
+- **SC-003**: Duplicate IDs, branch collisions, cycles, dirty repositories, and
+  malformed result contracts fail before unsafe continuation.
+- **SC-004**: Alternative consolidation cannot proceed without explicit
+  selection.
+- **SC-005**: A 13-worker local agent smoke set completes natively on the
+  authorized development Mac without remote writes.
+- **SC-006**: Secure CaseTracker Units 00 through 03 complete in all six MSL
+  repositories with no more than three workers active, using the same
+  explicitly authorized native development override.
+
+## Boundaries
+
+- Version 0.1.0 is experimental and opt-in.
+- No model name is prescribed.
+- Provider-specific merge behavior is supplied through a local runner profile.
+- The workspace standard matrix remains at seven presets until field validation
+  succeeds.
+- The repository owner's 2026-07-18 instruction explicitly overrides
+  Container-First only for the native macOS development smoke set and the
+  Secure CaseTracker field campaign for this preset. The override is not a
+  reusable exception and expires when the field findings are captured.
+- The repository owner's 2026-07-18 instruction also permits the development
+  field campaign to classify only GitHub Actions jobs rejected before their
+  first step with the known account-payment or spending-limit annotation as an
+  external provider `N/A`. Local and agent-side gates remain required. Any
+  technical CI failure, review finding, or different unavailable check remains
+  blocking.
+- Container-First remains binding for normal Secure Trader work and for any
+  later production, learner, or routine campaign.

specs/007-parallel-autonomous-run-governance/tasks.md

@@ -0,0 +1,17 @@
+# Tasks: Parallel Autonomous Run Governance
+
+- [x] T001 Define campaign, runner, state, result, and handoff schemas.
+- [x] T002 Implement campaign validation and DAG checks.
+- [x] T003 Implement isolated worktree creation and bounded scheduling.
+- [x] T004 Implement status, cooperative stop, resume, and reconciliation.
+- [x] T005 Implement all-ready consolidation and human alternative selection.
+- [x] T006 Add Bash entry points and PowerShell comment-based help.
+- [x] T007 Add commands, agent guidance, runbook, evidence, and retrospective.
+- [x] T008 Add deterministic fixtures for all topologies and failure modes.
+- [x] T009 Run 13 local real-agent smoke workers natively on the authorized
+  development Mac.
+- [x] T010 Publish protected experimental repository and v0.1.0.
+- [ ] T011 Run Secure CaseTracker Units 00-03 across six MSL repositories
+  natively under the explicit, temporary native and GitHub-Billing development
+  overrides.
+- [ ] T012 Promote field findings, v0.2.0, standard matrix, and catalog entry.

specs/spec-kit-preset-repos/README.md

@@ -1,7 +1,7 @@
 # Spec Kit Preset Repository Scaffolds
 
-This directory contains repo-shaped release candidates for the seven standard
-home-baseline-derived Spec Kit governance presets.
+This directory contains repo-shaped release candidates for the eight standard
+and optional home-baseline-derived Spec Kit governance presets.
 
 Each subdirectory is intended to become a standalone GitHub repository:
 
@@ -12,6 +12,7 @@ Each subdirectory is intended to become a standalone GitHub repository:
 - `spec-kit-preset-agent-parity-governance`
 - `spec-kit-preset-cross-platform-governance`
 - `spec-kit-preset-autonomous-run-governance`
+- `spec-kit-preset-parallel-autonomous-run-governance`
 
 The scaffolds intentionally do not contain nested `.git/` directories.
 Create standalone repositories outside this root workspace before
@@ -31,7 +32,7 @@ specify preset resolve plan-template
 specify preset resolve tasks-template
 ```
 
-For the full stacked smoke test, install all seven presets in their
+For the full stacked smoke test, install all eight presets in their
 recommended priority order:
 
 ```bash
@@ -42,6 +43,7 @@ specify preset add --dev /path/to/spec-kit-preset-a11y-governance --priority 40
 specify preset add --dev /path/to/spec-kit-preset-cross-platform-governance --priority 50
 specify preset add --dev /path/to/spec-kit-preset-agent-parity-governance --priority 60
 specify preset add --dev /path/to/spec-kit-preset-autonomous-run-governance --priority 70
+specify preset add --dev /path/to/spec-kit-preset-parallel-autonomous-run-governance --priority 80
 specify preset list
 specify preset resolve constitution-template
 ```
@@ -59,6 +61,7 @@ specify preset resolve cross-platform-agent-guidance-addendum-template
 specify preset resolve a11y-agent-guidance-addendum-template
 specify preset resolve agent-parity-guidance-addendum-template
 specify preset resolve autonomous-run-agent-guidance-addendum-template
+specify preset resolve parallel-autonomous-agent-guidance-addendum-template
 ```
 
 Each preset also registers three wrap commands (`speckit.specify`,
@@ -73,4 +76,5 @@ specify preset info cross-platform-governance
 specify preset info a11y-governance
 specify preset info agent-parity-governance
 specify preset info autonomous-run-governance
+specify preset info parallel-autonomous-run-governance
 ```

specs/spec-kit-preset-repos/spec-kit-preset-a11y-governance/README.md

@@ -1,25 +1,41 @@
 # A11Y Governance Preset
 
 Version: `0.4.0`
-Requires: `spec-kit >= 0.8.0` (uses the `wrap` and `append` composition
+Status: published, standard governance preset
+Priority: `40`
+Requires: Spec-Kit `>=0.8.0` (uses the `wrap` and `append` composition
 strategies introduced in 0.8.x).
 
-Purpose:
+## Zweck / Purpose
 
 - inject accessibility, bilingual-delivery, and CEFR-B2 readability
-  expectations into Spec Kit
+  expectations into Spec-Kit
 - make didactic inline-code-comment review explicit for new or changed
   non-trivial logic
 - preserve the `Programmierung #include<everyone>` principle as a
   reusable preset instead of a local-only policy
 
-Primary source chapters from `home-baseline` constitution:
+## Installation / Install
+
+```bash
+specify preset add \
+  --from https://github.com/hindermath/spec-kit-preset-a11y-governance/archive/refs/tags/v0.4.0.zip \
+  --priority 40
+specify preset info a11y-governance
+```
+
+```bash
+specify preset add --dev /path/to/a11y-governance --priority 40
+specify preset info a11y-governance
+```
+
+## Quellkapitel / Source Chapters
 
 - `VII. Programmierung #include<everyone> — Inclusion & Accessibility By
   Default`
 - `VIII. DE-First / EN-Second Bilingual Delivery`
 
-Standards and rules in scope:
+## Standards und Regeln / Standards and Rules
 
 - `WCAG 2.2 Level AA` baseline for every user-facing artefact
 - `DE first, EN second` delivery; bilingual `DE / EN` headings or a
@@ -34,7 +50,7 @@ Standards and rules in scope:
 - A11Y coverage for `CLI`, `documentation`, `HTML`, `UI`, and generated
   templates
 
-Preset strategy:
+## Preset-Strategie / Preset Strategy
 
 - append accessibility governance to `constitution-template`,
   `spec-template`, `plan-template`, and `tasks-template`
@@ -45,9 +61,10 @@ Preset strategy:
 - provide starter templates for A11Y review, bilingual content review,
   CLI accessibility review, and accessibility evidence
 
-Evidence templates included:
+## Evidenzvorlagen / Evidence Templates
 
-- Spec-Kit run evidence fields are embedded in the evidence templates to support audit-ready applicability, N/A rationale, reviewer, and follow-up records.
+- Spec-Kit run evidence fields are embedded in the evidence templates to support
+  audit-ready applicability, N/A rationale, reviewer, and follow-up records.
 - `a11y-checklist-template` (WCAG 2.2 AA criteria coverage)
 - `bilingual-content-check-template` (DE/EN headings, German
   orthography, CEFR-B2 readability, `*.EN.md` companion guidance)
@@ -59,7 +76,7 @@ Evidence templates included:
 
 Default evidence location: `docs/accessibility/`.
 
-When to use:
+## Einsatz / When to Use
 
 - any project that produces user-facing `CLI`, `documentation`, `HTML`,
   `UI`, or generated templates
@@ -68,22 +85,30 @@ When to use:
 - learning, training, or reference projects where non-trivial code logic
   should remain understandable for apprentices and future maintainers
 
-When not to use:
+## Nicht verwenden / When Not to Use
 
 - purely internal artefacts with no user-facing surface at all
 - teams that do not want DE-first / EN-second guidance
 
-Coverage note:
+## Sicherheit und Grenzen / Safety and Boundaries
+
+- Installation adds governance prompts, templates, and wrapped Spec-Kit
+  command guidance; it does not run accessibility tooling by itself.
+- The preset does not grant repository, remote, merge, deployment, or
+  provider-administration authority.
+- WCAG, bilingual, readability, and evidence decisions remain auditable
+  project decisions and must be recorded when declared `N/A`.
+
+## Abdeckung / Coverage
 
 - generated templates count as user-facing when humans are expected to
   read or edit them
 - CLI output, review checklists, and bilingual delivery all belong to
   the preset's scope
 
-Release notes:
-
-- `v0.4.0` adds audit-ready Spec-Kit run evidence fields so generated Markdown documents and checklists can record applicability, N/A rationale, reviewer, evidence path, residual risk, and follow-up per standards-relevant Spec-Kit run.
-
-Recommended standalone install priority:
+## Versionshinweise / Release Notes
 
-- `30`
+- `v0.4.0` adds audit-ready Spec-Kit run evidence fields so generated Markdown
+  documents and checklists can record applicability, N/A rationale, reviewer,
+  evidence path, residual risk, and follow-up per standards-relevant Spec-Kit
+  run.