claude-config

Global Claude Code configuration — agents, skills, plugins, and project templates.

Guide d'utilisation complet : voir USAGE.md — workflows typiques, exemples par type de projet (mobile, web, CLI, firmware, monorepo), arbre de décision "quel skill utiliser ?", cas de figure validés, et table des erreurs fréquentes.

---

Overview

This repo is your personal Claude Code setup, versioned and reproducible across machines.

claude-config/
├── CLAUDE.md              # Global coding preferences (style, rules, workflow)
├── settings.json          # Global permissions (100 deny / 18 ask / 57 allow rules)
├── install-plugins.sh     # One-shot installer: prerequisites + all plugins (reads plugins.lock.json)
├── link.sh                # Symlinks this repo into ~/.claude/
├── doctor.sh              # Setup diagnostic — checks symlinks, plugins, permissions, token budget
├── update-all.sh          # One-command update for all components
├── Makefile               # Unified entry point: make install / doctor / update
├── plugins.lock.json      # Version pinning for non-marketplace dependencies (RTK, GSD v2, ruflo)
├── version.txt            # Semver version of this config
├── CHANGELOG.md           # Release history
├── lib/
│   └── detect-plugins.sh  # Shared plugin detection — sourced by all scripts
├── hooks/
│   └── session-start.sh   # Health check + toggle plugin status at session start
├── skills-external/
│   └── gstack/            # Git submodule — garrytan/gstack (symlinked to ~/.claude/skills/gstack)
├── .gitmodules            # Submodule declaration
├── agents/
│   ├── analyzer.md        # Factual codebase analysis (read-only)
│   ├── interviewer.md     # Project questionnaire → PROJECT BRIEF
│   ├── onboarder.md       # Onboard existing project — CLAUDE.md, settings, optional GSD ROADMAP
│   ├── status-reporter.md  # Consolidated project status — read-only snapshot
│   ├── plugin-advisor.md  # Plugin check: detect signals, apply compatibility matrix, block if needed
│   ├── readme-updater.md  # Update README from git history + codebase
│   ├── refactorer.md      # Surgical refactoring with norm enforcement
│   └── scaffolder.md      # Full project generation (CLAUDE.md, README, code)
├── skills/
│   ├── analyze/           # /analyze — deep factual analysis
│   ├── health/            # /health — run setup diagnostic
│   ├── init-project/      # /init-project — full project initialization
│   ├── onboard/           # /onboard — onboard existing project into claude-config
│   ├── status/            # /status — consolidated project snapshot
│   ├── plugin-check/      # /plugin-check — check plugin config vs project needs
│   ├── readme/            # /readme — update README from current state
│   ├── refactor/          # /refactor — improve code without changing behavior
│   └── ship-feature/      # /ship-feature — ship a feature end-to-end
└── templates/
    ├── project-CLAUDE.md  # Template for per-project CLAUDE.md
    └── settings/
        ├── settings.json         # Template for project .claude/settings.json
        ├── settings.local.json   # Template for personal .claude/settings.local.json
        ├── .claudeignore         # Template for project .claudeignore
        └── SETTINGS.md           # Rule syntax reference (rule types, patterns, defaultMode values)

Architecture principle:

• skills/ = entry points you invoke via /skill-name
• agents/ = execution units called by skills (never invoked directly by user)
• lib/ = shared shell functions sourced by scripts (plugin detection)
• templates/ = symlinked to ~/.claude/templates/ — copy into projects via per-project setup
• Custom skills use Superpowers agents for implementation phases (required — auto-detected)
• Plugins (Superpowers, GStack, GSD v2, ruflo, etc.) install separately and complement custom skills

---

Fresh install (new machine)

# 1. Clone with submodules — choose any location
git clone --recurse-submodules git@github.com:youruser/claude-config.git
cd claude-config

# 2. Symlink into ~/.claude/
bash link.sh

# 3. Install prerequisites + all plugins (detects OS, reads pinned versions from plugins.lock.json)
bash install-plugins.sh

# 4. Context7 CLI (optional — for fast-evolving libs like Next.js, React, Prisma)
npm install -g ctx7
ctx7 setup --claude    # configures MCP + rules for Claude Code (OAuth login)
# Or use standalone: ctx7 docs /vercel/next.js "middleware"

# 5. Verify setup
bash doctor.sh

# 6. Restart Claude Code — plugins load automatically

All scripts use their own location to find the repo — run them from anywhere or from the repo directory. Symlinks point to the repo's actual path, so renaming or moving the repo requires re-running bash link.sh.

The install script handles: git, Node.js 22, Rust/Cargo, Python 3, RTK, GStack (submodule), GSD v2, and all marketplace plugins on Linux (apt/dnf/pacman) and macOS (brew).

RTK and GSD v2 versions are pinned in plugins.lock.json. The install script reads those versions automatically. Marketplace plugins install to ~/.claude/plugins/ (user scope).

Install output is logged to install-YYYYMMDD-HHMMSS.log in the repo directory for post-mortem debugging.

---

Available slash commands

Custom skills (this repo)

Command	Description
/analyze	Deep factual analysis of code before any modification
/refactor	Improve code quality without changing behavior (strict norms)
/readme	Full README audit — diff vs codebase, mandatory stop, surgical updates
/plugin-check	Check active plugins vs project needs — recommend enable/disable
/init-project	Initialize a complete project from scratch (full orchestrator)
/ship-feature	Ship a feature end-to-end with validation gates (full orchestrator)
/health	Run setup diagnostic — check symlinks, plugins, permissions, token budget
/onboard	Onboard an existing project — generate CLAUDE.md, settings, optional GSD v2 ROADMAP

Superpowers skills (auto-invoked or explicit)

Required dependency. Superpowers must be active for /init-project and /ship-feature. The plugin-advisor (STEP 0) blocks and shows install instructions if Superpowers is missing.

Command	When it auto-activates
superpowers:brainstorming	When you describe something to build
superpowers:writing-plans	After design is approved
superpowers:subagent-driven-development	With an approved plan
superpowers:requesting-code-review	Auto — after a feature step
superpowers:finishing-a-development-branch	After review is approved

GStack skills (Garry Tan — full-product projects only)

Installed as a git submodule at skills-external/gstack/, symlinked to ~/.claude/skills/gstack/. Use when: project has UI + design + deploy + browser QA. Skip for backend/lib/CLI projects. Full command reference: ~/.claude/skills/gstack/README.md or run /office-hours to start.

GSD v2 — standalone CLI (multi-session large features)

Architecture change from v1: GSD v2 (gsd-pi) is a standalone TypeScript CLI built on the Pi SDK. It is not a Claude Code plugin — it runs as an external process with its own session management. The /gsd ... commands are GSD-internal and are typed inside a gsd terminal session, not in Claude Code.

Install: npm install -g gsd-pi (pinned version in plugins.lock.json)

Use when: a feature spans multiple days/sessions, you need crash recovery, cost tracking per unit, parallel workers across milestones, or automatic context-fresh execution per task.

# Start a GSD session in your terminal (from your project directory)
gsd

# Or jump straight to autonomous mode — walk away and come back to built software
gsd          # then inside the session:
/gsd auto    # autonomous mode: research → plan → execute → commit → repeat
/gsd         # step mode: pause between each unit for review
/gsd status  # progress dashboard
/gsd discuss # talk through architecture decisions
/gsd quick   # atomic quick task without full planning overhead

Key commands inside a GSD session:

Command	Description
/gsd auto	Autonomous mode — research, plan, execute, commit, repeat until milestone done
/gsd or /gsd next	Step mode — execute one unit at a time, pause between each
/gsd quick	Quick atomic task with GSD guarantees (no full planning overhead)
/gsd stop	Stop auto mode gracefully
/gsd status	Progress dashboard (token usage, cost, milestone progress)
/gsd discuss	Discuss architecture decisions (works alongside auto mode)
/gsd steer	Hard-steer plan documents during execution
/gsd prefs	Model selection, timeouts, budget ceiling
/gsd doctor	Runtime health checks
/gsd migrate	Migrate a v1 .planning directory to .gsd format
/gsd export --html	Generate self-contained HTML report for a milestone
/worktree	Git worktree lifecycle — create, switch, merge, remove

GSD v2 vs v1:

	v1 (deprecated)	v2 (current)
Runtime	Claude Code slash commands	Standalone CLI (Pi SDK)
Context management	None — fills up	Fresh session per task
Auto mode	LLM self-loop	State machine with .gsd/ files
Crash recovery	None	Lock files + session forensics
Cost tracking	None	Per-unit token/cost ledger
Git strategy	LLM writes git commands	Worktree isolation, squash merge

Ruflo CLI (enterprise multi-agent orchestration)

Ruflo (formerly claude-flow) is a heavy enterprise multi-agent orchestration CLI — 310+ tools, 100+ agent types, WASM kernel, self-learning architecture. ~500-1500 tokens passive cost when hooks active.

Use when: project explicitly requires coordinating 5+ specialized agents simultaneously, parallel swarm execution, or enterprise-grade multi-agent orchestration. For standard multi-session work, GSD v2 is sufficient and much lighter.

Install:

# Minimal install (~40MB, no ML/embeddings)
npm install -g ruflo@latest --omit=optional

# Initialize in your project
ruflo init --wizard

# Useful commands
ruflo agent spawn -t coder     # spawn a coding agent
ruflo swarm init                # start a swarm
ruflo memory search -q "..."   # search vector memory
ruflo doctor                    # diagnostics

# Verify
ruflo --version

Bundled skills (Claude Code built-in, always available)

Command	Description
/batch <instruction>	Large-scale parallel refactoring — decomposes into 5–30 units, spawns one background agent per unit in isolated git worktrees
/debug [description]	Enable debug logging for the session, analyze the session debug log
/simplify [focus]	Review recent changes for code reuse, quality, efficiency issues

Other plugin commands

Command	Plugin	Description
/pr-review-toolkit:review-pr	pr-review-toolkit	Multi-agent PR review (6 specialized agents)
/context7:docs <lib>	context7	Manual doc lookup for a specific library (via ctx7 CLI)

---

Workflow patterns

Pattern A — Nouveau projet (court, 1 session)

/plugin-check "description"   → configure plugins
/init-project "description"   → interview → scaffold → implement v1
/ship-feature "feature"       → ship feature by feature

Pattern B — Nouveau projet (long, multi-session)

/plugin-check "description"
/init-project "description"   → à la fin, STEP 13 propose d'init GSD v2
                                 → répondre "yes"
# Ensuite dans un terminal :
gsd                            → démarrer une session GSD
/gsd auto                      → mode autonome — walk away
/gsd status                    → vérifier la progression
/gsd discuss                   → décisions d'architecture en cours de route

À chaque reprise de session :

/status                        → snapshot : plugins, token, git state, milestone GSD en cours

Pattern C — Projet existant (onboarding)

cd mon-projet-existant/
# Dans Claude Code :
/onboard                       → génère CLAUDE.md + settings + .claudeignore
                                 → optionnel : ROADMAP.md pour GSD v2
/status                        → confirmer que l'onboarding est complet + état du projet
/plugin-check "type de projet"
/ship-feature "prochaine feature"

Pattern D — Hotfix / modification ponctuelle

# Pas de /init-project, pas de GSD
/analyze src/module-cible.py   → rapport factuel sans solution
/ship-feature "corriger X"     → brainstorm + plan + gate + impl + review

Pattern E — Refactoring ciblé

/analyze src/legacy.py         → liste les violations
/refactor src/legacy.py        → corrections sans changement de comportement

Choisir entre /ship-feature et gsd auto

Critère	/ship-feature	gsd auto
Durée estimée	< 1 journée	> 1 journée
Nombre de tâches	< 10	> 10
Crash recovery nécessaire	non	oui
Suivi de coût par tâche	non	oui
Workers parallèles	non	oui (parallel mode)
Contexte fresh par tâche	non (même session)	oui (Pi SDK)

---

Orchestrators in detail

/init-project

Same rigor as /ship-feature. Two validation gates. Full TDD subagent pipeline for v1 features. The Scaffolder only creates the skeleton (no features, no README). readme-updater handles the README in two passes: CREATE then SYNC.

STEP 0 blocks if Superpowers is not installed (required for steps 3, 6, 8, 10, 11).

/init-project <project idea>
    │
    ├── STEP 0:  PLUGIN CHECK (plugin-advisor)        ← blocks if Superpowers missing or wrong plugins
    ├── STEP 1:  INTERVIEWER (custom)                 → PROJECT BRIEF
    ├── STEP 2:  ANALYZER (custom)                    → ANALYSIS REPORT
    ├── STEP 3:  superpowers:brainstorming             → VALIDATED DESIGN
    ├── STEP 4:  VALIDATION GATE #1                   → approve architecture
    ├── STEP 5:  SCAFFOLDER (custom)                  → skeleton only (CLAUDE.md +
    │                                                    settings + structure +
    │                                                    empty entry points, NO features,
    │                                                    NO README)
    ├── STEP 5b: README-UPDATER create mode (custom)  → CREATE README from CLAUDE.md
    ├── STEP 6:  superpowers:writing-plans             → decompose v1 features into tasks
    ├── STEP 7:  VALIDATION GATE #2                   → approve task plan
    ├── STEP 8:  superpowers:subagent-driven (TDD)    → implement each feature (isolated)
    ├── STEP 9:  ANALYZER (custom)                    → regression + deviation check
    ├── STEP 10: superpowers:requesting-review         → full code review
    ├── STEP 11: superpowers:finishing-branch          → cleanup + build + tests
    └── STEP 12: README-UPDATER sync mode (custom)    → sync README with implementation

/ship-feature

STEP 0 blocks if Superpowers is not installed (required for steps 1, 2, 4, 6, 7).

/ship-feature <feature description>
    │
    ├── STEP 0: PLUGIN CHECK (plugin-advisor) ← blocks if Superpowers missing or wrong plugins
    ├── STEP 1: superpowers:brainstorming     → VALIDATED DESIGN
    ├── STEP 2: superpowers:writing-plans     → task plan
    ├── STEP 3: VALIDATION GATE               → user approval required
    ├── STEP 4: superpowers:subagent-driven   → implementation (TDD)
    ├── STEP 5: ANALYZER (custom)             → regression / deviation check
    ├── STEP 6: superpowers:requesting-review → code review
    ├── STEP 7: superpowers:finishing-branch  → cleanup
    └── STEP 8: README-UPDATER sync mode      → sync README with new feature

/plugin-check

Standalone command you can run at any time to audit your plugin config against what you're about to do. Also embedded as STEP 0 in both orchestrators.

Blocks if Superpowers is not active (required by orchestrators). Blocks if critical project-specific plugins are missing (frontend tools, Context7, GStack). Warns if ruflo is active with no multi-agent signal, or if GSD v2 CLI is not installed for multi-session work.

/plugin-check "I want to build a React + FastAPI SaaS"

→ Detects active plugins
→ Scans filesystem for project signals (frontend? design? deploy? multi-agent?)
→ Applies compatibility matrix
→ Produces recommendation table with passive cost estimate
→ Warns about plugin conflicts (gstack + ruflo, etc.)
→ Blocks with OPTIONS if critical plugins are missing
→ Or confirms "proceed" if config is optimal

---

Plugin compatibility matrix

Quick reference

Pair	Relation	Notes
frontend-design ↔ ui-ux-pro-max	⚠️ Overlap	Keep both for design-heavy. Drop ui-ux-pro-max for simple UI.
gstack ↔ gsd v2	✅ Complementary	Different scopes — CC workflow vs CLI orchestration
gstack ↔ ruflo	⚠️ Overlap	Both orchestrate multi-step work. Use one or the other. ~3250-4250t combined.
gsd v2 ↔ ruflo	⚠️ Overlap	Sequential (GSD) vs parallel swarm (ruflo). Pick based on need.
superpowers ↔ gsd v2	✅ Complementary	Single-session engine + multi-session CLI = no conflict
superpowers ↔ gstack	✅ Complementary	Used together by orchestrators
context7 ↔ any	✅ Independent	Doc lookup CLI (ctx7) — always safe to combine

Recommended sets by project type

Project type	Plugins ON	OFF	Passive cost
Backend API / microservice	superpowers, context7*	frontend-design, ui-ux-pro-max, gstack, ruflo	~800t
Frontend SPA / SSR	superpowers, frontend-design, ui-ux-pro-max, context7	gstack, ruflo	~1600t
Full-stack SaaS	superpowers, gstack, frontend-design, ui-ux-pro-max, context7	ruflo	~4400t
CLI tool / library	superpowers	all toggles	~800t
Multi-session large feature	superpowers + gsd v2 CLI (external)	ruflo	~800t CC
Quick fix / hotfix	superpowers	all toggles	~800t
Design system / component lib	superpowers, frontend-design, ui-ux-pro-max	gstack, ruflo, gsd	~1600t
Enterprise multi-agent	superpowers, ruflo + gsd v2 CLI (external)	others	~2300t CC

*context7 only if using fast-evolving libs (Next.js, React 18+, Prisma, Supabase) security-guidance and rtk are ALWAYS ON (0 tokens) — omitted from estimates

---

Plugins reference

All plugins below are installed by install-plugins.sh.

How loading works

Only each skill's description field is pre-loaded into the system prompt at session start — the full skill body is loaded on demand when the skill is invoked. CLAUDE.md is the only file loaded in full at every session. Disabling a plugin prevents even its description from loading.

A hooks/session-start.sh hook shows plugin toggle status at every session start. Run /plugin-check anytime to get a recommendation for the current project type.

Plugin	Status	Passive cost	When to use	Installed by
security-guidance	✅ ALWAYS ON	0 tokens (hook only)	—	claude-code-plugins
RTK	✅ ALWAYS ON	0 tokens (hook only)	—	cargo (pinned in plugins.lock.json)
Superpowers	✅ REQUIRED	~600–1000 tokens	— required by orchestrators	superpowers-marketplace
GStack	🔄 TOGGLE	~2500–3000 tokens	Full-product: UI + design + deploy + browser QA	git submodule
GSD v2	🖥️ CLI	0 tokens (external CLI)	Multi-day features, crash recovery, cost tracking, parallel workers	npm (pinned in plugins.lock.json)
ruflo	🔄 TOGGLE	~500–1500 tokens	Enterprise multi-agent swarm (5+ concurrent agents)	npm (CLI)
plugin-dev	🔄 TOGGLE	~100 tokens	Creating plugins or custom skills	claude-code-plugins
pr-review-toolkit	🔄 TOGGLE	~300 tokens	PR review sessions	claude-code-plugins
frontend-design	🔄 TOGGLE	~200 tokens	Any project with a UI	claude-code-plugins
ui-ux-pro-max	🔄 TOGGLE	~400 tokens	Design system, color/typography choices	ui-ux-pro-max-skill
Context7 CLI	🔄 TOGGLE	~200 tokens	Fast-evolving libs (Next.js, React, Prisma…)	npm (ctx7) + optional ctx7 setup --claude

Rule: toggle plugins are OFF by default. /plugin-check signals when to enable them. If you use /init-project or /ship-feature, plugin-check runs automatically as STEP 0 and blocks if Superpowers is not active.

Marketplaces

Plugins are installed from GitHub-hosted marketplaces. Three are used by this config:

Marketplace	GitHub repo	Plugins	Auto-available
claude-plugins-official	anthropics/claude-plugins-official	Anthropic-curated third-party plugins	✅ yes
claude-code-plugins	anthropics/claude-code	Anthropic bundled plugins (security-guidance, frontend-design, pr-review-toolkit, plugin-dev)	❌ add manually
superpowers-marketplace	obra/superpowers-marketplace	Superpowers workflow plugin	❌ add manually
ui-ux-pro-max-skill	nextlevelbuilder/ui-ux-pro-max-skill	UI/UX Pro Max design plugin	❌ add manually

install-plugins.sh adds all required marketplaces automatically.

Manual install example:

# Add the marketplace (once)
claude plugin marketplace add anthropics/claude-code

# Install a plugin from it
claude plugin install --scope user frontend-design@claude-code-plugins

# Browse all available plugins
/plugin   # → Discover tab

Version pinning

RTK, GSD v2, and ruflo versions are pinned in plugins.lock.json:

{
  "rtk":   { "source": "https://github.com/rtk-ai/rtk", "version": "v0.34.3" },
  "gsd":   { "source": "npm:gsd-pi",  "version": "2.64.0" },
  "ruflo": { "source": "npm:ruflo",   "version": "3.5.58" }
}

install-plugins.sh reads these versions automatically. To update a pinned version: edit plugins.lock.json, then re-run install-plugins.sh. GStack is pinned via its git submodule pointer.

Disabling a plugin for a specific project

# In Claude Code
/plugin
# → Find the plugin → toggle off for this scope

Or in the project's .claude/settings.json:

{
  "enabledPlugins": {
    "gstack@gstack": false
  }
}

---

Settings and permissions

File hierarchy

Highest
  managed-settings.json        — enterprise, cannot be overridden
  CLI flags                    — --allowedTools / --disallowedTools (session only)
  .claude/settings.local.json  — personal machine overrides (gitignored)
  .claude/settings.json        — project rules (committed to project repo)
  ~/.claude/settings.json      — global user rules (this repo's settings.json)
Lowest

DENY always wins over ALLOW at any level.
.claudeignore applies independently of all permission rules.

Global settings (this repo's settings.json)

100 deny rules, 18 ask rules, 57 allow rules.

Section	Purpose
deny — secrets (Read)	Blocks Read on .env, .pem, .key, SSH keys, cloud credentials
deny — secrets (Bash)	Blocks cat, head, tail, grep, less, more on .env and secret files
deny — env leak	Blocks env, printenv, export * — prevents secret exposure via process environment
deny — secret move	Blocks cp/mv on .env* and secrets/ — closes copy-then-read bypass
deny — destructive	Blocks rm -rf, git push --force, chmod 777
deny — system	Blocks sudo, ssh, scp, crontab, systemctl
deny — injection	Blocks curl | bash, wget | sh
deny — escalation	Blocks bash -c, eval, exec, find -delete, perl -e, ruby -e
deny — runtime exec	Blocks python3 -c *, node -e *, source /dev/stdin, mkfifo *
deny — exfiltration	Blocks xargs * .env*, tar * .env*, zip * .env*, base64 .env*
ask — risky	Prompts before git push, docker run, package managers
ask — write tools	Prompts before xargs, all sed (including in-place)
ask — stash destructive	Prompts before git stash pop, drop, clear
allow — safe reads	Auto-approves git read-only, ls, cat, grep, find
allow — stash safe	Auto-approves git stash (push), list, show
disableBypassPermissionsMode	Prevents YOLO mode globally
disableAutoMode	Prevents auto mode globally

Per-project setup

cd your-project
mkdir -p .claude

# Find the repo from any existing symlink
CONF="$(dirname "$(readlink ~/.claude/CLAUDE.md)")"

# Project settings (commit to project git)
cp "$CONF/templates/settings/settings.json" .claude/settings.json

# Personal overrides (never commit — gitignore it)
cp "$CONF/templates/settings/settings.local.json" .claude/settings.local.json
echo ".claude/settings.local.json" >> .gitignore

# Hard file exclusions (commit to project git)
cp "$CONF/templates/settings/.claudeignore" .claudeignore

# Project CLAUDE.md (commit to project git)
cp "$CONF/templates/project-CLAUDE.md" CLAUDE.md

---

Updating

One-command update (recommended)

# From the repo directory
bash update-all.sh
# Pulls config, prompts before updating GStack (tracks main), updates RTK + GSD v2 (pinned),
# updates ruflo if installed, refreshes symlinks, runs doctor

Manual updates

This repo

git pull
# Symlinks → changes active immediately

GStack (submodule)

# Option A — inside Claude Code (recommended)
/gstack-upgrade

# Option B — via submodule (from the repo directory)
# Note: GStack tracks branch = main, review upstream commits before updating
git submodule update --remote skills-external/gstack
cd skills-external/gstack && ./setup
git add skills-external/gstack
git commit -m "chore: update gstack to latest"

RTK

# Uses the version pinned in plugins.lock.json
bash update-all.sh

# Or manually
cargo install --git https://github.com/rtk-ai/rtk --tag v0.34.3 --force

GSD v2

# Uses the version pinned in plugins.lock.json
bash update-all.sh

# Or manually
npm install -g gsd-pi@2.64.0

Ruflo CLI

# Uses the version pinned in plugins.lock.json
bash update-all.sh

# Or manually
npm install -g ruflo@3.5.58

Marketplace plugins

/plugin marketplace update    # inside Claude Code

---

Adding a new custom skill

Fastest way:

make new-skill name=myskill
# Creates agents/myskill.md + skills/myskill/SKILL.md with templates filled
# Edit both files, then: bash link.sh

Manually:

1. Create agents/myagent.md — role, tasks, rules, output format
2. Create skills/myskill/SKILL.md:

---
name: myskill
description: What this skill does — front-load key use case (max 250 chars)
argument-hint: <what to pass>
disable-model-invocation: true
allowed-tools: Read, Write, Edit, Bash, Grep, Glob
---

Load and follow strictly:
- $HOME/.claude/agents/myagent.md

Execute on:

$ARGUMENTS

3. Or use /plugin-dev:create-plugin to generate a skill from conversation.

---

Per-project agent overrides

Override any global agent for a specific project:

CONF="$(dirname "$(readlink ~/.claude/CLAUDE.md)")"
cp "$CONF/agents/refactorer.md" .claude/agents/refactorer.md
# Edit .claude/agents/refactorer.md — the local version takes precedence

---

Maintenance

Diagnostic

# Quick check from terminal (from the repo directory)
bash doctor.sh

# Or from within Claude Code
/health       # full diagnostic (symlinks, plugins, permissions, token budget)
/status       # project snapshot at session start (plugins, git, GSD milestone)

# Unified commands via Makefile (from the repo directory)
make doctor     # diagnostic
make update     # pull + submodules + symlinks + doctor
make install    # link.sh + install-plugins.sh
make onboard    # reminder to run /onboard in Claude Code
make new-skill name=myskill  # scaffold agent + skill files

doctor.sh checks 7 axes: symlinks, GStack submodule (with pinning warning), prerequisites (git, Node, Cargo, Python, Claude Code), plugins (RTK, Superpowers, Context7, GSD v2, ruflo), permissions (deny count, bypass mode), token budget (breakdown vs Pro session budget), and config consistency (frontmatter, CRLF detection).

session-start.sh runs a quick health check at every session start (filesystem only, no subprocesses) and displays toggle plugin status, GSD v2 CLI status, with /plugin-check and /health hints.

Both scripts source lib/detect-plugins.sh for consistent plugin detection logic.

Updating

# One-command update (from the repo directory)
bash update-all.sh

# Or step by step
git pull                                                # this repo
# GStack: prompts for confirmation (tracks main branch)
git submodule update --remote skills-external/gstack
bash link.sh                                            # refresh symlinks
bash doctor.sh                                          # verify

---

Troubleshooting

"command not found" after install

Restart your shell or run source ~/.bashrc / source ~/.zshrc.

Orchestrator blocks at STEP 0 — Superpowers missing

Install: claude plugin marketplace add obra/superpowers-marketplace && claude plugin install --scope user superpowers@superpowers-marketplace Then re-run the orchestrator.

"agent not found" or hallucinated agent content

Symlinks are broken. cd into your config repo and run bash link.sh, then verify with bash doctor.sh.

GStack skills not showing up

Run bash link.sh and verify: ls -la ~/.claude/skills/gstack. If missing: cd into your config repo and run git submodule update --init.

GStack submodule "directory not found after init"

The submodule is not registered in .gitmodules (never added). Fix:

cd ~/Documents/claude   # chemin de ton config repo
git submodule add https://github.com/garrytan/gstack skills-external/gstack
git submodule update --init --recursive
bash link.sh
git add .gitmodules skills-external/gstack && git commit -m "chore: add gstack submodule"

link.sh warns "is a real directory"

If ~/.claude/agents/, ~/.claude/skills/, ~/.claude/lib/, or ~/.claude/templates/ exist as real directories, the script skips them to avoid data loss. Rename or remove the directory, then re-run link.sh.

GSD v2 — "command not found: gsd"

npm's global bin directory is not in $PATH. Run npm prefix -g to find it, then add $(npm prefix -g)/bin to your PATH. See the GSD troubleshooting guide.

GSD v2 — migrating from v1 projects

If you have old projects with .planning directories from GSD v1, migrate them:

cd your-project
gsd          # start a session
/gsd migrate # migrate .planning → .gsd format

Ruflo CLI not detected by doctor.sh

Ruflo must be installed globally. Run:

npm install -g ruflo@latest --omit=optional
ruflo --version

Token budget exceeded — skills truncated at session start

Too many plugins active. Run /plugin-check to optimize. Run bash doctor.sh for a token budget breakdown (vs Pro ~11k session budget).

settings.json not applying

Check precedence: deny always wins over allow at any level. .claudeignore overrides all permission rules. Verify deny count: cat ~/.claude/settings.json | python3 -c "import json,sys; print(len(json.load(sys.stdin)['permissions']['deny']))" Expected: 100 deny rules.

Claude reads .env despite deny rules

The Read(**/.env) deny rule blocks the Read tool. Bash(cat .env) and similar commands have separate deny rules (included in this config). For hard exclusion regardless of tool, use .claudeignore.

install-plugins.sh failed — where are the logs?

Check install-YYYYMMDD-HHMMSS.log in your config repo directory.

---

Known limitations

• Deny rules are pattern-based, not sandboxed. Core bypass vectors (bash -c, eval, python3 -c *, node -e *, source /dev/stdin, mkfifo *, xargs * .env*, base64 .env*) are blocked. Process substitution (<(cmd), >(cmd)), here strings (<<<), and /dev/fd/* access remain possible without explicit patterns — .claudeignore is the only hard file exclusion mechanism.
• disableAutoMode syntax not verified against CC v2.1.89 — added as "disableAutoMode": "disable" by analogy with disableBypassPermissionsMode. # TODO: VERIFY
• Superpowers is a hard dependency for /init-project and /ship-feature. The plugin-advisor (STEP 0) auto-detects and blocks if missing, with install instructions. No manual fallback mode.
• Marketplace plugin versions are not pinned. They install latest. Non-marketplace tools (RTK, GSD v2, ruflo) are pinned in plugins.lock.json.
• Token budget: CLAUDE.md loads in full every session (~420t). Skill bodies load on-demand. Plugin descriptions load passively. With all toggles active, passive plugin cost can reach ~50% of the Pro session budget (~11k tokens/5h). Run /health or bash doctor.sh for a breakdown.
• GSD v2 is a standalone CLI, not a Claude Code plugin. /gsd ... commands are GSD-internal and do not work in the Claude Code slash command bar.
• Ruflo is heavy (~340MB default, ~500-1500t passive tokens). Only enable for genuine enterprise multi-agent needs. For multi-session work, GSD v2 is lighter and sufficient.
• Agent frontmatter fields model, memory, effort are enforced by Claude Code v2.1.x.
• bypassPermissions mode is disabled via disableBypassPermissionsMode.
• GStack submodule is pinned to branch = main, not a commit hash. update-all.sh now prompts for confirmation before updating. Review upstream commits before accepting.
• disable-model-invocation: true on orchestrator skills (/init-project, /ship-feature): behavior when a skill with this flag invokes sub-agents via loaded agent files has not been fully verified in CC v2.1.89. # TODO: VERIFY