claude-platform-reference
/claude-platform-reference
Podaje szczegóły konfiguracji platformy Claude Code w SpawnForge: hooki, skille, serwery MCP i wtyczki.
- Autor
- Tristan578 ↗
- Repozytorium
- project-forge ↗
- Licencja
- Apache-2.0
- Ostatnia zmiana repozytorium
- 17.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 1
Brak sygnałów. Repozytorium zmienione 17.10.2026, 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
Podaje szczegóły konfiguracji platformy Claude Code w SpawnForge: hooki, skille, serwery MCP i wtyczki.
Oryginalny opis (EN)
Use when you need details about SpawnForge's Claude Code platform setup — hooks, skills inventory, MCP servers, validation tools, or plugin config. Reference for the automation infrastructure.
Jak działa
- Skills (ls .claude/skills/ for the live count)
- MCP Servers (.mcp.json — 8 servers)
- Hooks (.claude/hooks/ — 52 scripts, 20 event types)
- Validation Tools (.claude/tools/)
- Plugin (.claude-plugin/plugin.json)
- Approved Specs
Podgląd SKILL.md
--- name: claude-platform-reference description: "Use when you need details about SpawnForge's Claude Code platform setup — hooks, skills inventory, MCP servers, validation tools, or plugin config. Reference for the automation infrastructure." --- # Claude Code Platform Reference ## Skills (`ls .claude/skills/` for the live count) **Orchestration:** `/planner`, `/builder`, `/cycle` **Engine:** `/rust-engine`, `/build`, `/arch-validator`, `/rust-best-practices` **Web:** `/frontend`, `/mcp-commands`, `/web-accessibility`, `/shadcn`, `/vercel-react-best-practices`, `/vercel-composition-patterns`, `/vercel-react-view-transitions` **Next.js:** `/next-best-practices`, `/next-cache-components`, `/next-upgrade` **Testing:** `/testing`, `/playwright-best-practices`, `/tdd` **Infrastructure:** `/infra-services`, `/troubleshoot`, `/kanban`, `/babysit-prs`, `/pr-code-review`, `/pr-green-machine`, `/env-health-check`, `/changelog-review`, `/deploy-to-vercel`, `/resolve-pr-comments`, `/resolve-all-pr-comments`, `/autonomous-sprint` **Database:** `/db-migrate`, `/neon-postgres`, `/claimable-postgres`, `/neon-postgres-egress-optimizer` **Deployment:** `/deploy-engine` (user-only) **Billing:** `/stripe-webhooks` (background, auto-loads on billing edits) **Features:** `/game-engine`, `/game-ui-design`, `/multiplayer-readiness`, `/viewport` **Workflow:** `/design`, `/architect-flow`, `/docs`, `/developer-experience` **Review:** `/review-protocol`, `/component-checklist` **Vercel:** `/vercel-cli-with-tokens` ## MCP Servers (`.mcp.json` — 8 servers) - `context7` — live library documentation for all 30+ dependencies - `neon` — direct Neon Postgres queries (needs `NEON_API_KEY`) - `playwright` — browser automation for E2E verification - `github` — GitHub API access for PR/issue/Actions operations - `sentry` — error tracking, issue search, event analysis - `stripe` — payment processing, webhook debugging, test cards - `upstash` — Redis cache operations and rate limiting - `taskboard` — the local ticket board (launched through a git alias; see `docs/guides/taskboard-sync.md`) Claude Code defers MCP tools behind Tool Search by default. Only `context7` sets `"alwaysLoad": true`, so its tools are in context from session start: it needs no credential, only reads documentation, and nearly every session uses it. Every other server stays deferred on purpose. `neon`, `sentry`, `stripe` and `upstash` read credentials from the environment, and `github` needs a GitHub token and can write to the repository (push, merge, create and delete files), so loading its tools into every context, reviewer seats included, is not worth the saved search. `alwaysLoad` also makes startup wait for the server. Add it to another server only if that server needs no credential, cannot write, and is needed in nearly every session. The key and its behavior are documented at https://code.claude.com/docs/en/mcp#exempt-a-server-from-deferral. The key affects Claude Code only: Codex does not read `.mcp.json`. ## Hooks (`.claude/hooks/` — 52 scripts, 20 event types) **Event types:** PreToolUse, PostToolUse, SessionStart, SessionEnd, UserPromptSubmit, Stop, SubagentStart, SubagentStop, TaskCreated, TaskCompleted, TeammateIdle, PreCompact, PostCompact, StopFailure, ConfigChange, InstructionsLoaded, WorktreeCreate, CwdChanged, PostToolUseFailure, FileChanged | Hook | Event | Purpose | |------|-------|---------| | `inject-lessons-learned.sh` | PreToolUse (Edit/Write/Bash) | Anti-patterns before action | | `pre-push-quality-gate.sh` | PreToolUse (git push) | Lint+tsc before push | | `block-main-commits.sh` | PreToolUse (git commit) | Prevents commits on main | | `verify-branch.sh` | PreToolUse (Edit/Write) | Prevents edits on wrong branch | | `check-db-transaction.sh` | PreToolUse (Edit/Write) | Warns on db.transaction() | | `review-quality-gate.sh` | Stop (reviewers) | Validates PASS/FAIL verdict | | `block-writes.sh` | PreToolUse (reviewers) | Reviewers are read-only | | `builder-quality-gate.sh` | Stop (builder) | Verifies lint+tests ran | | `cargo-check-wasm.sh` | PostToolUse (rust-engine) | Cargo check after .rs edits | | `validate-agent-output.sh` | SubagentStop | Verify reviewer output | | `reject-incomplete-review.sh` | TeammateIdle | Exit 2 if no verdict | | `save-critical-context.sh` | PreCompact | Dump decisions before compaction | | `restore-context-hints.sh` | PostCompact | Inject context after compaction | | `on-session-start.sh` | SessionStart | Taskboard state + sync | | `on-stop.sh` | Stop / SessionEnd | Worktree safety commit + GitHub sync | ## Validation Tools (`.claude/tools/`) - `validate-rust.sh` — Architecture boundaries, bridge isolation, unsafe audit - `validate-frontend.sh` — ESLint, TypeScript (web, packages/ui, apps/docs), vitest - `validate-mcp.sh` — Manifest sync, MCP type-check + tests, AI parity audit - `validate-tests.sh` — Test inventory, coverage report - `validate-docs.sh` — Documentation integrity, version refs - `dx-audit.sh` — Cross-IDE consistency, tool health, onboarding check - `validate-all.sh` — Run all validators ## Plugin (`.claude-plugin/plugin.json`) Internal plugin manifest. Test with `claude --plugin-dir .` ## Approved Specs - `specs/2026-03-25-game-creation-orchestrator-phase2a-v4.md` — Game Creation Orchestrator (systems-not-genres)
Wersje
W bazie nie ma innych skilli o nazwie claude-platform-reference.
Identyczna treść występuje jeszcze w 1 innych folderach na GitHubie. Tu pokazujemy jedną, najlepiej ocenioną kopię.
Migawka bazy z 13.11.2026 · wersja paczki 92486322a4e6 (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
- 100working-with-claude-code
Dostarcza dokumentację Claude Code: CLI, wtyczki, hooki, serwery MCP, skille i konfigurację. - 100validate-all
Uruchamia pełną walidację claudelint na plikach projektu Claude Code: CLAUDE.md, skille, hooki i serwery MCP. - 100claude-code-plugin-development
Prowadzi przez tworzenie pluginów Claude Code, które pakują agentów, hooki, komendy, skille lub serwery MCP w dystrybuowalną całość. - 100claude-automation-recommender
Analizuje bazę kodu i rekomenduje automatyzacje Claude Code: hooki, subagenty, skille, pluginy i serwery MCP, przy optymalizacji konfiguracji. - 100claude-automation-recommender
Analizuje bazę kodu i rekomenduje automatyzacje Claude Code: hooki, subagenty, skille, pluginy i serwery MCP, przy optymalizacji konfiguracji. - 100claude-automation-recommender
Analizuje bazę kodu i rekomenduje automatyzacje Claude Code: hooki, subagenty, skille, pluginy i serwery MCP, przy optymalizacji konfiguracji.