configuration-and-credentials
/configuration-and-credentials
Przegląda i zmienia ustawienia środowiska Open Agentic Catalog, aktywację możliwości oraz rotację sekretów i poświadczeń.
- Repozytorium
- open-agentic-catalog ↗
- Licencja
- Apache-2.0
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- 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.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
Co robi i kiedy
Przegląda i zmienia ustawienia środowiska Open Agentic Catalog, aktywację możliwości oraz rotację sekretów i poświadczeń.
Oryginalny opis (EN)
Review or change Open Agentic Catalog runtime settings, capability activation, secret delivery and credential rotation. Use for precedence, validation, loading and replacement semantics; image mechanics and local startup use build-and-deployment, and chart implementation uses helm-and-minikube.
Jak działa
- Trace delivery and activation
- Change the smallest owning contract
- Verify the consuming boundary
Podgląd SKILL.md
--- name: configuration-and-credentials description: Review or change Open Agentic Catalog runtime settings, capability activation, secret delivery and credential rotation. Use for precedence, validation, loading and replacement semantics; image mechanics and local startup use build-and-deployment, and chart implementation uses helm-and-minikube. --- # Configuration and credentials Work from the repository root. Resolve instruction paths from that root. Read `AGENTS.md`, the configuration reference (`documentation/src/content/docs/reference/configuration.mdx`) and only the affected backend loader (`internal/platform/config/README.md`), sample profiles (`config/README.md`) or independent UI host (`catalog-ui/internal/web/README.md`). Preserve review-only and unit-only scope. Changing code or documenting rotation does not authorize replacing credentials in a retained deployment. ## Trace delivery and activation Identify the setting's consuming role, non-secret configuration owner, secret/trust input where applicable, and loading point. Trace defaults, selected JSON, allowlisted environment overrides, Make exports, Compose inputs and Helm templates only where they participate. Binaries do not load env files; Make's selected workflow file is executable Make syntax. `CATALOG_ENV_FILE=/dev/null` omits that file, not inherited overrides. Chart-pinned values and the independent UI's strict JSON loader have distinct precedence; backend overrides do not configure the UI. Distinguish delivery, process activation and upstream acceptance/revocation. Use the per-consumer rotation table (`documentation/src/content/docs/operations/maintenance.mdx#rotate-credentials-and-trust`) rather than assume mounted file changes reload a client. Most inputs are startup-loaded; source credential bindings resolve complete immutable generations per request. Existing Secret/ConfigMap content changes alone do not establish a Helm rollout. Preserve direct/file mutual exclusion, bounded secret parsing, safe diagnostics and separate API, worker, owner and UI authority. The three live tenant settings (`documentation/src/content/docs/reference/configuration.mdx#live-tenant-settings`) are retained application policy under ADR-0025 (`documentation/src/content/docs/architecture/decisions/0025-tenant-operational-settings.md`). Trace their typed API/store and admission consumers instead of adding loader, environment or chart fields. Preserve inheritance, current permissions, exact ETags, atomic audit, policy-read failure and unchanged admitted work. Deployment changes do not overwrite saved overrides; diagnostics show startup inputs, not active tenant policy. ## Change the smallest owning contract For a setting change, update typed decoding/defaults, cross-field and role validation, the explicit override allowlist, safe reports, affected profiles and configuration documentation together. Update `.env.example` and actual Compose/Helm delivery when relevant. Include omission, invalid/null/duplicate/unknown input, role applicability and meaningful enabled/disabled controls. Keep unavailable capabilities unavailable; a feature flag or provider declaration cannot implement them. Lower limits prospectively without rewriting retained evidence or replay responses. For activation or rotation, follow the specific runbook: UI session keys use staged reader/writer overlap; cursor keys have no overlap ring; PostgreSQL password acceptance must be tested with new connections; source binding generations are replaced atomically; webhook signing rotation and deployment ciphertext-key replacement are different operations. Webhook maintenance (`documentation/src/content/docs/operations/webhooks.mdx#deployment-encryption-key-replacement`) owns its stopped-runtime and backup-key requirements. Select only the relevant procedure, preserve recovery material, and never create a blanket hot-reload or restart controller. External secret managers deliver existing scoped inputs. The owned Minikube scenario qualifies selected Vault/VSO delivery, rotation and recovery. Follow the maintenance guide for tested scope and remaining limits. Vault remains optional and is not a Catalog dependency. Keep real secrets out of profiles, values, fixtures, reports and command output. Use security review (`.agents/skills/security-review/SKILL.md`) for substantive secret/trust-boundary changes and recovery and repair (`.agents/skills/recovery-and-repair/SKILL.md`) when retained ciphertext or restored state is involved. ## Verify the consuming boundary Use `make config-check` with explicit `CATALOG_CONFIG_FILE` and `CATALOG_CONFIG_ROLE` for static backend validation. It loads the files required by that role's static contract but does not contact dependencies or qualify database privileges, credential acceptance or UI configuration. It executes the CLI and is not a pure-unit gate; under an explicit unit-only limit, use isolated loader/host tests and report command/deployment qualification as unrun. Check its documented validation layers before choosing a deeper diagnostic. Reuse unit testing (`.agents/skills/unit-testing/SKILL.md`) for parser/loading/privacy decisions and the existing functional fixtures (`.agents/skills/functional-testing/SKILL.md`) for real replacement, retired-input rejection and retained-state checks. Use build and deployment (`.agents/skills/build-and-deployment/SKILL.md`) for launcher delivery and Helm and Minikube (`.agents/skills/helm-and-minikube/SKILL.md`) for rendered settings, mount contracts and affected rollouts. Select applicable checks through post-code verification (`.agents/skills/post-code-verification/SKILL.md`). Report which consumer loaded which generation using safe identity fields, actual acceptance tests, restart/overlap requirements and unrun boundaries. Update the configuration reference and owning maintenance procedure rather than duplicate another input inventory here.
Wersje
W bazie nie ma innych skilli o nazwie configuration-and-credentials.
Migawka bazy z 13.11.2026 · wersja paczki 07e90059c581 (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
- 85helm-and-minikube
Przegląda, aktualizuje i testuje wykresy Helm dla Open Agentic Catalog oraz wdraża i diagnozuje scenariusze w Minikube. - 95build-and-deployment
Utrzymuje budowanie i wdrażanie Open Agentic Catalog: obrazy, Compose oraz lokalne środowisko API, interfejsu i dokumentacji. - 100kubernetes-deployment
Obsługuje operacje na poziomie klastra Kubernetes: cykl życia Minikube i k3s, reguły zapory GCP, debugowanie minikube tunnel i wzorce sekretów startowych. - 100secrets-config-audit
Sprawdza gotowość sekretów i konfiguracji wdrożeniowej bez ujawniania wartości: zmienne środowiskowe, mapy konfiguracji i rotację. - 98provider-configuration
Implementuje konfigurację i uwierzytelnianie providera Terraform w Plugin Framework: schemat poświadczeń, zmienne środowiskowe, łańcuch dostawców poświadczeń, ochrona sekretów i walidacja przy konfiguracji. - 90helm-chart-quality-review
Przegląda jakość, bezpieczeństwo i testowalność chartu Helm, w tym Chart.yaml, values.yaml, templates, tests i konfigurację CI chart-testing.