Skip to content

CLAUDE.md

Latest commit

 

History

History
24 lines (15 loc) · 1015 Bytes

CLAUDE.md

File metadata and controls

24 lines (15 loc) · 1015 Bytes
Raw

CLAUDE.md

Claude Code entry point. All project rules, skill routing, and workflow live in AGENTS.md.

@AGENTS.md

Claude Code Specific

Slash Commands

.claude/commands/*.md are symlinks to .agents/skills/*/SKILL.md. Available: /branch-review, /create-issue, /create-pr, /logic-design.

Skill Auto-Matching

Natural-language triggers auto-invoke skills via description frontmatter matching. Examples: "PR 올려줘" → create-pr, "리뷰해줘" → branch-review.

CRITICAL — Agent Usage Policy

You MUST NEVER spawn or invoke any agent without explicit user approval.

  • Before using any agent, always discuss the task with the user first. Do not proceed with agent execution on your own — present the plan, explain which agent will be used and why, and wait for explicit approval.
  • If the user has already granted approval upfront, proceed with the agent work immediately. Once the work is complete, provide a clear bulleted list of everything that was done.