release-planning
/release-planning
Planuje wydanie lub hotfix: zakres wdrożenia, bramki wydania oraz zapis autoryzacji produkcyjnej i wyjątków zatwierdzonych w czacie.
- Autor
- arozumenko ↗
- Repozytorium
- sdlc-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 3
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 13.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Planuje wydanie lub hotfix: zakres wdrożenia, bramki wydania oraz zapis autoryzacji produkcyjnej i wyjątków zatwierdzonych w czacie.
Oryginalny opis (EN)
Use when preparing a release or hotfix, changing deployment scope, defining release gates, or recording production authorization and chat-approved exceptions.
Jak działa
- Plan
- Authorization and chat overrides
- Record checkpoints
- Completion
Podgląd SKILL.md
--- name: release-planning description: Use when preparing a release or hotfix, changing deployment scope, defining release gates, or recording production authorization and chat-approved exceptions. --- # Release planning Produce an executable plan and durable record for the actual artifact and target. This skill owns the shared release authority and evidence contract used by pre-production-checks, progressive-delivery, release-recovery and infrastructure-monitoring. ## Plan 1. Read existing release policy, workflow, deployment configuration and open incidents. Identify service, source commit, immutable artifact digest/version, configuration revision and exact target account, region, cluster/namespace or non-Kubernetes equivalent. Inspect installed tool versions. Load [platform guidance](references/platforms.md) only for the detected stack. 2. Use the project's release-record location or `docs/releases/<release-id>/`. Start from [the record template](templates/release-record.md). Reference restricted logs securely; keep secrets and production personal data out of Git. 3. Map dependency order, migration compatibility, capacity and blast radius with developers and Tech Lead. Select rolling, canary, blue/green or another existing strategy. State why it fits the workload, traffic and available routing. 4. Define stages, observation windows, minimum sample counts, fresh metric queries, thresholds, abort conditions, smoke tests and ongoing monitoring owner. Low traffic needs an explicit alternative, not invented statistical confidence. 5. Specify the known-good artifact, recovery mechanism, compatibility evidence, expected recovery time, data-loss implications and authorization. Coordinate application, configuration, infrastructure and data changes; a container rollback does not reverse a migration. Route unresolved architecture to Tech Lead. 6. Record environment execution authority, pipeline concurrency mechanism, and continuation owner. Hand the plan to pre-production-checks. Planning does not deploy; emergency changes retain explicit checks or recorded exceptions. ## Authorization and chat overrides Use existing policy and prior session approvals. Where authority is absent, finish preparation and ask for the specific production action. Read-only investigation and reversible local preparation need no extra release approval. A missing, stale, failed or inconclusive required gate holds promotion by default. Explain the concrete risk and options. **Explicit written human approval in chat can waive that gate**; no separate approval system or repeated confirmation is required. “Yes” accepts a clearly scoped exception just presented. A generic “ship it” without a disclosed gap is not automatically an informed waiver. Record the named approving human (never invent identity). If the host cannot supply identity, ask for the approver's name in chat and record it as self-declared. Record timestamp/message reference, exact approval, release/artifact/configuration/target, waived gates, disclosed risks, allowed actions and any time/stage limits. Carry this record across skills and resumed sessions. Proceed without asking again while approval covers the action; re-open only for materially different risks or changed scope. Keep gate status **waived**, never passed. A destructive action always needs a specific human confirmation for the current situation, even when a runbook preauthorizes it. Distinguish gate waiver from deployment/remediation authority; the same chat message can grant both. Approval cannot bypass credentials, enforced access controls or technical impossibility. When the record cannot be written to the repo, carry it in chat marked "persistence pending", keep it visibly unfinished, and write it to the repo at the first opportunity. A release cannot be closed while the record is still pending persistence. Do not erase approval or require the human to repeat it solely because the record store is unavailable. ## Record checkpoints Create append-only checkpoints on a dedicated release-record branch, committed and pushed, recording approvals, run/lock state and the next observation owner. Handoffs cite the branch and exact commit so a resumed session can continue before merge. Tech Lead reviews and PM merges through the normal gate. An unmerged checkpoint never counts as completion. ## Completion Return record path, artifact/target, strategy, gate and recovery readiness, authority/exception references, unresolved decisions and next owner. Keep merged, deployed, verified healthy and closed with accepted risk distinct. Closed with accepted risk requires a named human's explicit waiver of the remaining observation. Never report it as verified healthy. The record must list the evidence gaps and the follow-up owner.
Pliki w paczce (3)
- SKILL.md 5 KB
- references/platforms.md 4 KB
- templates/release-record.md 4 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · arozumenko/sdlc-skills | 90 | MIT |
| qa-aman/claude-skills | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 48410901a2c5 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100hotfix-release
Tworzy gałąź wydania hotfix, robi cherry-pick commitów, podnosi wersję i otwiera wstępny pull request, na przykład dla paczek NuGet i npm oraz backportów poprawek. - 90release-checklist
Przygotowuje rutynowe wydanie po starcie: zakres z gotowych historyjek, status CI, bezpieczeństwo migracji, flagi, changelog i kroki wdrożenia oraz wycofania. - 100qa-release
Prowadzi bramki wydania QA: decyzje go lub no-go, raporty podpisu wydania, triage defektów i komunikację gotowości do wdrożenia. - 100emergency-hotfix-deployment-procedure
Definiuje proces wdrożenia awaryjnej poprawki produkcyjnej poza zwykłym cyklem wydań: które kontrole można bezpiecznie pominąć, wymagane zatwierdzenie i plan wycofania. - 90deciding-release-scope
Ustala zakres wydania podczas planowania, tak aby wynik był identyfikowalny, niezależnie sprawdzalny i bezpieczny do przekazania innemu agentowi. - 90release
Przygotowuje wersjonowane wydanie: zakres, artefakty, zgodność, notatki i gotowość; wdrożenie na środowisko to osobny krok.