{"id":47630326,"url":"https://github.com/louisschlegel/claudekit","last_synced_at":"2026-04-01T23:21:25.174Z","repository":{"id":345326745,"uuid":"1185450085","full_name":"louisschlegel/claudekit","owner":"louisschlegel","description":"Universal Claude Code template — auto-configuring, self-improving, with 14 agents, 14 workflows, and full-stack coverage","archived":false,"fork":false,"pushed_at":"2026-03-18T17:19:14.000Z","size":232,"stargazers_count":0,"open_issues_count":2,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-03-19T06:06:11.745Z","etag":null,"topics":["ai","automation","boilerplate","claude","claude-code","developer-tools","devops","template"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/louisschlegel.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":".github/CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-03-18T15:41:43.000Z","updated_at":"2026-03-18T17:17:46.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/louisschlegel/claudekit","commit_stats":null,"previous_names":["louisschlegel/claudekit"],"tags_count":5,"template":false,"template_full_name":null,"purl":"pkg:github/louisschlegel/claudekit","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/louisschlegel%2Fclaudekit","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/louisschlegel%2Fclaudekit/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/louisschlegel%2Fclaudekit/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/louisschlegel%2Fclaudekit/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/louisschlegel","download_url":"https://codeload.github.com/louisschlegel/claudekit/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/louisschlegel%2Fclaudekit/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31292881,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-01T21:15:39.731Z","status":"ssl_error","status_checked_at":"2026-04-01T21:15:34.046Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","automation","boilerplate","claude","claude-code","developer-tools","devops","template"],"created_at":"2026-04-01T23:21:24.222Z","updated_at":"2026-04-01T23:21:25.165Z","avatar_url":"https://github.com/louisschlegel.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# claudekit\n\n[![Validate](https://github.com/louisschlegel/claudekit/actions/workflows/validate.yml/badge.svg)](https://github.com/louisschlegel/claudekit/actions/workflows/validate.yml)\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)\n[![Version](https://img.shields.io/badge/version-1.4.0-blue.svg)](.template/version.json)\n\nUn template auto-configurant, auto-améliorant et critique par design, qui génère toute l'infrastructure Claude Code adaptée à n'importe quel projet — 25 hooks, 23 agents, 36 workflows, 24 skills, 11 commands, 7 rules, config generator, injection defense, auto-update — à partir d'un simple interview conversationnel.\n\n---\n\n## Installation\n\n### Nouveau projet\n```bash\ngit clone https://github.com/louisschlegel/claudekit mon-projet\ncd mon-projet\nclaude \"setup claudekit\"\n```\n\n### Projet existant — one-liner\n```bash\ncd mon-projet/\ncurl -fsSL https://raw.githubusercontent.com/louisschlegel/claudekit/main/install.sh | bash\nclaude \"setup claudekit\"\n```\n\n\u003e **Important :** lancez `claude \"setup claudekit\"` (avec le message entre guillemets), pas juste `claude`. Le message initial déclenche l'interview de configuration automatiquement. Sans message, Claude attend votre input.\n\nClaude détecte votre stack, présente ce qu'il a trouvé, puis pose les questions de configuration une par une → génère toute la config adaptée à votre projet.\n\n---\n\n## Ce que tu obtiens après le setup\n\n```\nton-projet/\n├── CLAUDE.md                    # Orchestrateur — routing 32 intents\n├── project.manifest.json        # Config générée lors du setup\n├── learning.md                  # Mémoire institutionnelle (auto-alimentée)\n├── .mcp.json                    # Serveurs MCP configurés\n├── .claude/\n│   ├── settings.local.json      # Permissions Bash adaptées à ton stack\n│   ├── hooks/                   # 25 hooks\n│   ├── agents/                  # 23 agents (YAML frontmatter: tools, model, memory)\n│   ├── skills/                  # 24 skills (TDD, premortem, code-review, configure, etc.)\n│   ├── commands/                # 11 commands (/check-security, /new-module, etc.)\n│   └── rules/                   # 7 path-scoped rules (security, testing, critical-thinking)\n├── workflows/                   # 36 workflows end-to-end\n├── .claude-plugin/plugin.json   # Plugin system compatibility\n└── examples/claude-md/          # 4 example CLAUDE.md (Next.js, Django, Go, Rust)\n```\n\n---\n\n## Architecture du système\n\n```\ntemplate/\n├── CLAUDE.md                        # Orchestrateur — routing table (32 intents) + règles\n├── project.manifest.json            # {} → déclenche setup | rempli → contexte\n├── project.manifest.EXAMPLE.json   # Référence complète du schema\n├── learning.md.template             # Template pour la mémoire persistante\n├── install.sh                       # One-liner pour projets legacy\n│\n├── scripts/\n│   ├── gen.py                       # Générateur principal (manifest → tout) — --dry-run, --diff\n│   ├── claudekit.py                 # CLI unifié : validate, check, gen, bump, status, install\n│   ├── migrate-template.py          # Migration manifest entre versions (1.0.x → 1.1.0 → ...)\n│   ├── auto-learn.py                # Extrait les handoffs JSON agents → learning.md (--deduplicate)\n│   ├── self-improve.py              # Moteur d'observation (log friction events)\n│   ├── version-bump.py              # Semantic versioning (patch/minor/major)\n│   └── changelog-gen.py             # git log → CHANGELOG.md (conventional commits)\n│\n├── examples/\n│   ├── web-app.manifest.json        # SaaS Next.js + FastAPI + PostgreSQL\n│   ├── api.manifest.json            # REST API Python + Docker\n│   ├── ml.manifest.json             # MLOps PyTorch + MLflow + FastAPI\n│   └── mobile.manifest.json         # React Native + Expo + Turborepo\n│\n├── .claude/\n│   ├── hooks/\n│   │   ├── session-start.sh         # Bootstrap (40+ détecteurs stack + 12 signaux)\n│   │   └── user-prompt-submit.sh    # Bootstrap (32 intents + 8 patterns injection)\n│   └── agents/                      # 23 agents spécialisés\n│       ├── architect.md             # Conception + ADR + JSON handoff\n│       ├── reviewer.md              # Code review (BLOCKER/WARNING/SUGGESTION)\n│       ├── tester.md                # Tests exhaustifs\n│       ├── security-auditor.md      # OWASP + deps + secrets\n│       ├── debug-detective.md       # Root cause + 5 whys + git bisect\n│       ├── deployer.md              # Déploiement sécurisé + rollback\n│       ├── explorer.md              # Cartographie de codebase\n│       ├── doc-writer.md            # Docstrings + README + API docs\n│       ├── performance-analyst.md   # Profiling + optimisation\n│       ├── release-manager.md       # Orchestration release + changelog\n│       ├── cost-analyst.md          # Optimisation coûts cloud (AWS/GCP/Azure) + LLM tokens\n│       ├── spec-reader.md           # Parse cahier des charges → manifest + backlog + issues\n│       ├── data-engineer.md         # Pipelines data, dbt, Airflow, streaming\n│       ├── ml-engineer.md           # MLOps : train → serve → monitor\n│       ├── devops-engineer.md       # Infra, CI/CD, observabilité, résilience\n│       └── template-improver.md     # Meta-agent : améliore le template\n│\n├── workflows/                       # 31 workflows end-to-end\n│   ├── feature.md                   # Feature branch → merge + auto-learn\n│   ├── bugfix.md                    # Bug → root cause → test → fix\n│   ├── hotfix.md                    # Correctif urgent prod (express)\n│   ├── release.md                   # Tests → audit → changelog → tag → deploy\n│   ├── security-audit.md            # Scan complet + findings\n│   ├── refactor.md                  # Refactoring sécurisé\n│   ├── onboarding.md                # Nouveau projet ou legacy\n│   ├── self-improve.md              # Auto-amélioration du template\n│   ├── dependency-update.md         # Audit → grouper par risque → update\n│   ├── db-migration.md              # Migrations SQL zero-downtime + rollback\n│   ├── incident-response.md         # P1/P2 — mitigation → RCA → post-mortem\n│   ├── performance-baseline.md      # Profiling → baseline → benchmark\n│   ├── publish-package.md           # npm/PyPI/crates.io — build → sign → publish\n│   ├── api-design.md                # API-first : spec → review → mock → implem\n│   ├── a-b-test.md                  # Power analysis → feature flags → significance → ship/kill\n│   ├── data-quality.md              # Great Expectations + dbt + ISO 8000 score + SLA\n│   ├── llm-eval.md                  # RAGAS + hallucination detection + BLEU/ROUGE + deploy gate\n│   ├── spec-to-project.md           # Cahier des charges → manifest + backlog + arch + GitHub issues\n│   ├── code-review.md               # PR review structuré (BLOCKER/WARNING/SUGGESTION) → gh pr review\n│   ├── monitoring-setup.md          # Prometheus/Grafana/Loki/Sentry → dashboards + alertes + SLOs\n│   ├── cost-optimization.md         # Audit cloud + LLM → recommandations ROI + budget alerts\n│   └── dependency-audit.md          # CVE + licences + deps fantômes → rapport sans modification\n│\n└── .template/\n    ├── version.json                 # Version du template + historique\n    ├── known-patterns.json          # Patterns en attente de validation\n    └── improvements.log             # Observations JSONL (runtime, non commité)\n```\n\n---\n\n## Comment ça marche\n\n### 1. Intent classification (UserPromptSubmit hook)\n\nChaque message est analysé avant d'arriver à Claude :\n- **Injection detection** : bloque 50+ patterns de prompt injection (5 catégories)\n- **Intent classification** : 32 intents détectés automatiquement\n- **Output injection defense** : scanne les résultats Read/Bash/WebFetch pour les injections\n\n| Intent | Mots-clés déclencheurs |\n|--------|----------------------|\n| `hotfix` | urgence prod, emergency, production down |\n| `incident` | incident prod, outage, SLA breach, P1/P2 |\n| `db-migration` | migration db, alter table, add column |\n| `perf-test` | load test, benchmark, locust, k6 |\n| `publish` | publie sur npm/pypi, npm publish |\n| `api-design` | design api, nouvel endpoint, openapi |\n| `ab-test` | a/b test, feature flag, expérience, power analysis, significance |\n| `data-quality` | qualité des données, great expectations, dbt test, validation données |\n| `llm-eval` | évalue le rag, llm eval, ragas, hallucination, benchmark llm |\n| `spec-to-project` | cahier des charges, voici les specs, voici mon brief, PRD, analyse ce document |\n| `code-review` | review cette PR, relis ce code, code review, review le diff, relecture |\n| `monitoring-setup` | setup monitoring, prometheus, grafana, datadog, observabilité, alertes |\n| `cost-optimization` | optimise les coûts, trop cher AWS, facture cloud, rightsizing, coûts LLM |\n| `dependency-audit` | audit les dépendances, vérifie les CVE, npm audit, pip-audit, licence check |\n| `feature` | implémente, ajoute, nouvelle feature |\n| `bugfix` | bug, crash, erreur, fixe, regression |\n| `release` | release, prépare une version, tag v |\n| `security-audit` | audit, sécurité, vulnérabilité, CVE |\n| `update-deps` | mets à jour les dépendances, upgrade packages |\n| `refactor` | refactor, nettoie, restructure |\n| `onboard` | setup, initialise, configure le projet |\n| `improve-template` | améliore le template, self-improve |\n| `question` | comment, explique, pourquoi, what is |\n\n### 2. Contexte session (SessionStart hook)\n\n**Manifest vide** → détection automatique du stack (40+ marqueurs) : package.json, requirements.txt, go.mod, Cargo.toml, dbt_project.yml, MLproject, pubspec.yaml, build.gradle, *.tf, serverless.yml, etc.\n\n**Manifest rempli** → 12 signaux opérationnels injectés à chaque session :\n\n| Signal | Source |\n|--------|--------|\n| Couverture de tests | coverage.xml / lcov / coverage-summary.json |\n| Migrations en attente | `python3 manage.py makemigrations --check` |\n| Dépendances vulnérables | pip-audit / npm audit |\n| Branches stale (\u003e 14j) | git for-each-ref |\n| Statut CI/CD | `gh run list` |\n| Fichiers chauds sans tests | git log --name-only + analyse |\n| Dette technique | grep TODO/FIXME/HACK |\n\n### 3. Sécurité pré-exécution (PreToolUse)\n\nBloque avant exécution : `rm -rf /`, `DROP DATABASE`, `git push --force origin main`, `chmod 777`, etc.\n\n### 4. Qualité post-edit (PostToolUse)\n\nAprès chaque fichier modifié (selon `guards` dans le manifest) :\n- **lint** : ruff / eslint\n- **type_check** : tsc --noEmit\n- **migration_check** : Django makemigrations --check\n- **i18n_check** : parité clés fr/en\n- **test_on_edit** : pytest/jest sur le fichier correspondant\n\n### 5. Auto-learning (Stop hook + auto-learn.py)\n\nChaque fin de session → observation JSONL dans `.template/improvements.log` → tous les N sessions → agent `template-improver` analyse et améliore.\n\n`auto-learn.py` extrait les HANDOFF JSON des agents → sections structurées dans `learning.md`.\n\n---\n\n## Réseau d'agents\n\nTous les agents ont un **HANDOFF JSON structuré** pour passage de contexte, et des **checklists par type de projet** (web-app, api, mobile, library, data, ml, iac).\n\n| Agent | Rôle | Écrit |\n|-------|------|-------|\n| `architect` | Conception + ADR + trade-offs | learning.md |\n| `reviewer` | Code review (BLOCKER/WARNING/SUGGESTION) | — |\n| `tester` | Tests exhaustifs + edge cases | Fichiers de test |\n| `security-auditor` | OWASP + deps + secrets | — |\n| `debug-detective` | Root cause + 5 whys + git bisect | — |\n| `deployer` | Déploiement + rollback | Tags git |\n| `explorer` | Cartographie codebase | — |\n| `doc-writer` | Docstrings + README + API docs | .md + docstrings |\n| `performance-analyst` | Profiling + optimisation | — |\n| `release-manager` | Orchestration release + changelog | CHANGELOG |\n| `cost-analyst` | Optimisation coûts cloud (AWS/GCP/Azure) + LLM tokens | — |\n| `spec-reader` | Parse cahier des charges → manifest + backlog + issues GitHub | project.manifest.json, learning.md, backlog.md |\n| `data-engineer` | Pipelines data, dbt, Airflow, streaming | dbt models, DAGs |\n| `ml-engineer` | MLOps : framing → train → serve → monitor | Scripts ML |\n| `devops-engineer` | Infra, CI/CD, observabilité, résilience | Dockerfile, CI, IaC |\n| `template-improver` | Améliore le template lui-même | Fichiers template |\n| `memory-curator` | Consolide la mémoire institutionnelle | learning.md |\n| `compliance-officer` | RGPD, SOC2, PCI-DSS, HIPAA compliance | — |\n| `ai-engineer` | LLM APIs, RAG, embeddings, model serving | — |\n| `realtime-architect` | WebSockets, SSE, event sourcing, CQRS | — |\n| `data-modeler` | Schema design, migrations, indexation | — |\n| `schema-designer` | OpenAPI, GraphQL, Protobuf, contract-first | — |\n| `devils-advocate` | Challenge decisions, find risks, alternatives | — |\n\n---\n\n## Workflows complets\n\n| Workflow | Intent | Agents impliqués |\n|----------|--------|-----------------|\n| `feature.md` | `feature` | architect → tester → reviewer → doc-writer |\n| `bugfix.md` | `bugfix` | debug-detective → tester → reviewer |\n| `hotfix.md` | `hotfix` | debug-detective → reviewer → deployer |\n| `release.md` | `release` | security-auditor → release-manager → deployer |\n| `security-audit.md` | `security-audit` | security-auditor → reviewer |\n| `refactor.md` | `refactor` | explorer → architect → tester → reviewer |\n| `onboarding.md` | `onboard` | explorer → architect → doc-writer |\n| `self-improve.md` | `improve-template` | template-improver |\n| `dependency-update.md` | `update-deps` | security-auditor → reviewer |\n| `db-migration.md` | `db-migration` | architect → tester → deployer |\n| `incident-response.md` | `incident` | debug-detective → devops-engineer |\n| `performance-baseline.md` | `perf-test` | performance-analyst → architect |\n| `publish-package.md` | `publish` | security-auditor → release-manager |\n| `api-design.md` | `api-design` | architect → reviewer → doc-writer |\n| `a-b-test.md` | `ab-test` | ml-engineer → tester → reviewer |\n| `data-quality.md` | `data-quality` | data-engineer → tester → reviewer |\n| `llm-eval.md` | `llm-eval` | ml-engineer → tester → reviewer |\n| `spec-to-project.md` | `spec-to-project` | spec-reader → architect → doc-writer |\n| `code-review.md` | `code-review` | explorer → security-auditor → reviewer → architect |\n| `monitoring-setup.md` | `monitoring-setup` | devops-engineer → architect → doc-writer |\n| `cost-optimization.md` | `cost-optimization` | cost-analyst → architect → devops-engineer |\n| `dependency-audit.md` | `dependency-audit` | security-auditor → reviewer |\n\n---\n\n## Examples de manifests\n\nLe dossier [`examples/`](examples/) contient des manifests pré-configurés :\n\n| Fichier | Stack |\n|---------|-------|\n| [`web-app.manifest.json`](examples/web-app.manifest.json) | Next.js + FastAPI + PostgreSQL + Vercel |\n| [`api.manifest.json`](examples/api.manifest.json) | FastAPI + PostgreSQL + Docker |\n| [`ml.manifest.json`](examples/ml.manifest.json) | PyTorch + MLflow + FastAPI |\n| [`mobile.manifest.json`](examples/mobile.manifest.json) | React Native + Expo + Turborepo |\n| [`iac.manifest.json`](examples/iac.manifest.json) | Terraform + AWS + EKS + Kubernetes |\n| [`cli.manifest.json`](examples/cli.manifest.json) | Python CLI (Typer + Rich) → PyPI |\n\nLe dossier [`examples/claude-md/`](examples/claude-md/) contient des CLAUDE.md pré-écrits pour des stacks spécifiques :\n- `nextjs-saas.md` — Next.js + Supabase + Stripe + Vercel\n- `django-api.md` — Django + DRF + Celery + PostgreSQL\n- `go-microservice.md` — Go + gRPC + PostgreSQL + Kubernetes\n- `rust-api.md` — Rust + Axum + SQLx + PostgreSQL\n\n---\n\n## Génération\n\n`scripts/gen.py` lit le manifest et génère :\n\n| Fichier | Ce qui est généré |\n|---------|-------------------|\n| `.claude/settings.local.json` | Permissions Bash adaptées au stack |\n| `.claude/hooks/session-start.sh` | Contexte + 12 signaux opérationnels |\n| `.claude/hooks/user-prompt-submit.sh` | 32 intents + injection detection |\n| `.git/hooks/pre-push` | Secret scan + lint + tests avant chaque push (installé par gen.py) |\n| `scripts/migrate-template.py` | Migration automatique manifest entre versions |\n| `.claude/hooks/pre-bash-guard.sh` | Blocage commandes destructives |\n| `.claude/hooks/post-edit.sh` | Guards qualité |\n| `.claude/hooks/stop.sh` | Auto-learning async |\n| `.mcp.json` | Serveurs MCP |\n\n---\n\n## Self-improvement\n\n```\nSession → stop.sh → self-improve.py (observation JSONL)\n                           ↓ tous les N sessions\n                  template-improver agent\n                           ↓\n    AUTO → appliqué | APPROVAL → diff proposé | PR → branche créée\n                           ↓\n              version-bump.py → .template/version.json\n```\n\n---\n\n## 6 couches de sécurité indépendantes\n\n1. **Permissions whitelist** — seules les commandes nécessaires au stack sont autorisées\n2. **Pre-tool gate** — bloque les commandes destructives (5 catégories, case-insensitive)\n3. **Post-edit guards** — lint + type-check + scan secrets après chaque édition\n4. **Prompt injection detection** — bloque les tentatives de manipulation dans les messages\n5. **Output injection defense** — scanne les résultats Read/Bash/WebFetch pour les injections (50+ patterns)\n6. **Permission auto-approval** — auto-approve les read-only, bloque les accès sensibles (.env, .ssh, /etc/)\n\n---\n\n## Stacks supportés\n\n**Langages (31) :** Python, TypeScript, JavaScript, Go, Rust, Ruby, PHP, Java, Elixir, Swift, Kotlin, C, C++, C#/.NET, F#, Dart/Flutter, Scala, Clojure, Groovy, R, Bash, Lua, Perl, Haskell, Nim, Zig, OCaml\n\n**Frameworks web :** Django, FastAPI, Flask, React, Next.js, Vue, Nuxt, Angular, Svelte, Remix, Astro, NestJS, Express, Laravel, Spring Boot, Quarkus\n\n**Data/ML :** PyTorch, TensorFlow, JAX, scikit-learn, XGBoost, LightGBM, Hugging Face, LangChain, LlamaIndex, MLflow, W\u0026B, DVC, dbt, Airflow, Prefect, Dagster, Spark, Ray, Kafka, FAISS, Chroma, Qdrant, vLLM, Ollama\n\n**Mobile/Desktop :** Flutter, React Native, Expo, Electron, Tauri, iOS (Swift), Android (Kotlin)\n\n**Infra :** Docker, Kubernetes, Terraform, Ansible, Vercel, Railway, AWS, GCloud, Azure, Serverless, Firebase, Cloudflare Workers\n\n**Bases de données :** PostgreSQL, MySQL, Redis, MongoDB, SQLite, Elasticsearch, Cassandra, DynamoDB, BigQuery, Snowflake, ClickHouse, Neo4j, Pinecone\n\n**Package managers :** npm, pnpm, yarn, bun, deno, pip, uv, poetry, conda, cargo, nix\n\n**Monorepos :** Turborepo, Nx, Lerna, Bazel, Buck, Pants\n\n---\n\n## MCP Servers configurables\n\n| Server | Usage | Prérequis |\n|--------|-------|-----------|\n| `filesystem` | Accès fichiers étendu | npx |\n| `github` | Issues, PRs, code search | `GITHUB_TOKEN` |\n| `postgres` | Requêtes directes | `DATABASE_URL` |\n| `sqlite` | Base SQLite locale | npx |\n| `brave-search` | Recherche web | `BRAVE_API_KEY` |\n| `slack` | Notifications | `SLACK_BOT_TOKEN` |\n| `linear` | Issues Linear | `LINEAR_API_KEY` |\n| `notion` | Pages Notion | `NOTION_API_KEY` |\n| `playwright` | Browser automation | npx |\n| `desktop-commander` | Contrôle desktop | npx |\n\n\u003e Gmail, Google Calendar et Canva sont des intégrations natives Claude.ai — pas de configuration MCP nécessaire.\n\n---\n\n## Portabilité\n\nTous les hooks utilisent `$BASH_SOURCE` pour se localiser — le template fonctionne depuis n'importe quel emplacement, sans configuration.\n\n```bash\nSCRIPT_DIR=\"$(cd \"$(dirname \"${BASH_SOURCE[0]}\")\" \u0026\u0026 pwd)\"\nPROJECT_ROOT=\"$(dirname \"$(dirname \"$SCRIPT_DIR\")\")\"\n```\n\n---\n\n## Commandes utiles\n\n```bash\n# claudekit CLI (nouveau en v1.1.0)\npython3 scripts/claudekit.py status     # Vue d'ensemble : version, agents, workflows, scripts\npython3 scripts/claudekit.py validate   # CI complète en local\npython3 scripts/claudekit.py gen        # Régénère la config depuis project.manifest.json\npython3 scripts/claudekit.py bump patch # Bump version (patch/minor/major)\npython3 scripts/claudekit.py install /path/to/project  # Installe dans un projet existant\n\n# gen.py options avancées\npython3 scripts/gen.py --dry-run        # Prévisualise les fichiers générés sans écrire\npython3 scripts/gen.py --diff           # Montre les diffs avec la config actuelle\n\n# auto-learn\npython3 scripts/auto-learn.py --deduplicate  # Déduplique les entrées cross-session\n\n# migrate-template (mise à jour du template)\npython3 scripts/migrate-template.py --check  # Vérifie si des migrations sont disponibles\npython3 scripts/migrate-template.py          # Applique les migrations automatiquement\npython3 scripts/migrate-template.py --dry-run  # Simule sans écrire\n\n# make shortcuts\nmake validate      # CI complète en local\nmake check         # Validation rapide avant de pusher\nmake gen           # Alias pour gen.py\nmake bump-patch    # Bump version patch (1.0.0 → 1.0.1)\nmake changelog     # Met à jour CHANGELOG.md depuis git log\n```\n\n---\n\n## Documentation\n\n| Document | Description |\n|----------|-------------|\n| [`docs/manifest-reference.md`](docs/manifest-reference.md) | Référence complète du schema manifest (tous les champs, types, valeurs) |\n| [`CONTRIBUTING.md`](.github/CONTRIBUTING.md) | Ajouter un agent, un workflow, un stack |\n| [`SECURITY.md`](SECURITY.md) | Responsible disclosure + design sécurité |\n| [`examples/`](examples/) | Manifests pré-configurés pour 4 types de projet |\n\n---\n\n## Skills \u0026 Commands\n\n### Skills (invocables par Claude ou manuellement)\n\n| Skill | Description |\n|-------|-------------|\n| `/code-review` | 9 agents en parallèle → verdict Ready/Needs Attention/Needs Work |\n| `/tdd` | Red-Green-Refactor TDD enforcement |\n| `/premortem` | Imaginer l'échec → prévenir les risques |\n| `/review-architecture` | Analyse de trade-offs composant par composant |\n| `/handoff` | Sauvegarde structurée du contexte session |\n| `/configure` | Modifier le manifest en langage naturel |\n| `/auto-research` | Self-improving skill eval loop |\n| `/memory-sync` | Promouvoir learning.md → custom_rules |\n| `/update-claudekit` | Vérifier et appliquer les mises à jour |\n| `/changelog-tracker` | Suivre les updates Claude Code |\n\n### Commands (playbooks ad-hoc)\n\n| Command | Description |\n|---------|-------------|\n| `/generate-adr` | Architecture Decision Record generator |\n| `/check-security` | Audit de sécurité rapide |\n| `/new-module` | Scaffold adapté au stack |\n| `/new-entity` | Générer model + migration + CRUD + tests |\n| `/check-backend` | Health check backend |\n| `/incident-response` | Triage → diagnose → fix → post-mortem |\n| `/deploy-checklist` | Checklist pré-déploiement |\n| `/cost-report` | Rapport d'utilisation tokens/coûts |\n| `/setup-rules` | Auto-générer .claude/rules/ |\n| `/context-status` | État du contexte et recommandations |\n| `/secret-rotation` | Rotation de secrets en production |\n\n---\n\n## Auto-update\n\nclaudekit vérifie automatiquement les mises à jour à chaque session (cache 24h). Si une nouvelle version est disponible :\n\n```\n📦 claudekit 1.4.0 disponible (actuel: 1.3.1). Utilise /update-claudekit pour mettre à jour.\n```\n\nLes fichiers custom (manifest, learning.md, agents perso, workflows perso) sont **toujours préservés**.\n\n---\n\n## Contributing\n\nVoir [CONTRIBUTING.md](.github/CONTRIBUTING.md) — bugs, nouveaux agents, stacks, workflows.\n\nIdées et discussions : [GitHub Discussions](https://github.com/louisschlegel/claudekit/discussions)\n\n## License\n\n[MIT](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flouisschlegel%2Fclaudekit","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flouisschlegel%2Fclaudekit","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flouisschlegel%2Fclaudekit/lists"}