init
Analyze project structure and populate the "Repository Identity" section of CLAUDE.md (Zone B) plus AGENTS.md.
Best use case
init is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Analyze project structure and populate the "Repository Identity" section of CLAUDE.md (Zone B) plus AGENTS.md.
Teams using init should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/init/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How init Compares
| Feature / Agent | init | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Analyze project structure and populate the "Repository Identity" section of CLAUDE.md (Zone B) plus AGENTS.md.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
Related Guides
SKILL.md Source
# Initialize Project Configuration
Analyze this project and populate the **Repository Identity (Zone B)** section of `CLAUDE.md`, and mirror the same information into the project-specific sections of `AGENTS.md`.
## CLAUDE.md 3-zone layout (recap)
```
Zone A — Orchestra concept & template base (template-owned)
# @orchestra:template-boundary
Zone B — Repository Identity ← this skill writes here
# @orchestra:repo-boundary
Zone C — Working state (sessions, features, design pointers)
```
## Important
- Touch ONLY Zone B of `CLAUDE.md`. Never modify Zone A (above `@orchestra:template-boundary`) or Zone C (below `@orchestra:repo-boundary`).
- For `AGENTS.md`, do NOT modify the "Extensions" section and below — only update the top project-specific sections.
- If `CLAUDE.md` lacks the 3-zone markers (legacy layout), ask the user to run `./scripts/update.sh` first. Do not hand-insert markers.
## Steps
### 1. Project Analysis
Find these files to identify the tech stack:
- `package.json` → Node.js/TypeScript project
- `pyproject.toml` / `setup.py` / `requirements.txt` → Python project
- `Cargo.toml` → Rust project
- `go.mod` → Go project
- `Makefile` / `Dockerfile` → Build/deploy config
- `.github/workflows/` → CI/CD config
Also detect:
- npm scripts / poe tasks / make targets → Common commands
- Major libraries/frameworks
### 2. Ask User
Use AskUserQuestion tool to ask:
1. **Project overview**: What does this project do? (1-2 sentences)
2. **Code language**: English or Japanese for comments/variable names?
3. **Additional rules**: Any other coding conventions to follow?
### 3. Update CLAUDE.md Zone B
First verify the 3-zone markers exist:
```bash
grep -q "@orchestra:template-boundary" CLAUDE.md && grep -q "@orchestra:repo-boundary" CLAUDE.md
```
If either marker is missing, stop and ask the user to run `./scripts/update.sh` to migrate the file; the updater auto-migrates legacy single-boundary layouts.
Replace the content **between** the two markers with the following template (keep the marker lines and their ━ separators intact). Use the Edit tool by anchoring on the full block between the two boundary box lines.
```markdown
## Repository Identity
<!-- Managed by /init. Re-run /init to refresh. -->
### Project Overview
{User's answer to "what does this project do"}
### Language Settings
- **Thinking/Reasoning**: English
- **Code**: {Based on analysis — English or Japanese}
- **User Communication**: Japanese
### Tech Stack
- **Language**: {Detected language}
- **Package Manager**: {Detected tools}
- **Dev Tools**: {Detected tools}
- **Main Libraries**: {Detected libraries}
### Common Commands
```bash
{npm run dev / poe test / make build etc.}
```
```
### 4. Partial Update of AGENTS.md
Mirror the same information into `AGENTS.md` so Codex and Gemini see it. Update only the top section (up to the first `---`) with this format:
```markdown
# Project Overview
{User's answer}
## Language Settings
- **Thinking/Reasoning**: English
- **Code**: {Based on analysis - English or Japanese}
- **User Communication**: Japanese
## Tech Stack
- **Language**: {Detected language}
- **Package Manager**: {Detected tools}
- **Dev Tools**: {Detected tools}
- **Main Libraries**: {Detected libraries}
## Common Commands
```bash
{Detected commands}
```
```
### 5. Check Unnecessary Rules
Check rules in `.claude/rules/` and suggest removing unnecessary ones:
- Non-Python project → `dev-environment.md` (uv/ruff/ty) may not be needed
- No-test project → `testing.md` may not be needed
### 6. Report Completion
Report to user (in Japanese):
- Detected tech stack
- Files updated (`CLAUDE.md` Zone B, `AGENTS.md`)
- Recommended rules to remove (if any)Related Skills
context-loader
ALWAYS activate this skill at the start of every task. Load project context from .claude/ directory including coding rules, design decisions, and documentation before executing any task.
update-lib-docs
Update library documentation in .claude/docs/libraries/ with latest information from web search.
update-design
Explicitly update DESIGN.md with decisions from the current conversation. Use when you want to force a design document update.
troubleshoot
Diagnose and plan fixes for errors/bugs with Codex-first multi-agent collaboration (Codex + Opus 4.6 + Agent Teams). Codex CLI is consulted in EVERY phase for deep code reasoning, hypothesis evaluation, and fix validation. Phase 1: Error reproduction & context gathering (Opus subagent 1M context + Codex initial analysis + Claude user interaction). Phase 2: Parallel diagnosis (Agent Teams: Root Cause Analyst [Codex-driven] + Impact Investigator [Opus + Codex risk analysis]). Phase 3: Fix plan synthesis, Codex validation & user approval. Fix implementation is handled separately by /team-implement.
team-review
Parallel code review using Agent Teams. Spawns specialized reviewers (security, quality, test coverage) to review implementation from different perspectives simultaneously. Run after implementation.
team-implement
Parallel implementation using Agent Teams. Spawns teammates per module/layer, each owning separate files to avoid conflicts. Uses shared task list with dependencies for autonomous coordination. Run after /start-feature plan approval.
tdd
Implement features using Test-Driven Development (TDD) with Red-Green-Refactor cycle.
start-feature
Start a new feature with multi-agent collaboration (Opus 4.6 + Agent Teams). Phase 1: Codebase understanding (Opus subagent 1M context + Claude user interaction). Phase 2: Parallel research & design (Agent Teams: Researcher + Architect). Phase 3: Plan synthesis & user approval. Implementation is handled separately by /team-implement.
spike
Time-boxed technical investigation/feasibility study with Codex-first multi-agent collaboration (Codex + Opus 4.6 + Agent Teams). Codex CLI is consulted in EVERY phase for question framing, feasibility analysis, and final evaluation. Phase 1: Frame the investigation question & constraints (Claude user interaction + Codex question decomposition). Phase 2: Parallel investigation (Agent Teams: Researcher [Opus external research] + Feasibility Analyst [Codex deep analysis] + optional prototype). Phase 3: Codex synthesis into go/no-go recommendation & research report. Produces a DECISION DOCUMENT, NOT an implementation plan. Use /add-feature or /start-feature after a GO decision.
simplify
Simplify and refactor code while preserving functionality and library constraints.
research-lib
Research a library and create comprehensive documentation in .claude/docs/libraries/.
plan
Create a detailed implementation plan for a feature or task. Use when user wants to plan before coding.