skills.rest
lldwb
lldwbCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

repo-init

Generates AGENTS.md agent guidance and a CLAUDE.md pointer for git repositories.

1|Updated Sep 11, 2026
Search Tags:#claude code#claude.md#documentation generation#agents.md#agent guidance#repository onboarding#git verification
One-click install
npx skills add https://github.com/lldwb/lldwb-claude-skills --skill repo-init-lldwb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: repo-init
Source: https://github.com/lldwb/lldwb-claude-skills/tree/main/skills/repo-init
Command: npx skills add https://github.com/lldwb/lldwb-claude-skills --skill repo-init-lldwb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? New repositories lack accurate onboarding documentation for AI agents, and hand-written guidance often contains unverified claims, fabricated sections, or overwrites existing content. This Skill produces verified, merge-safe agent guidance grounded in actual repository facts. ## Core Features & Use Cases - Verified guidance generation: Explores README, build files, CI config, and existing AI rules, then verifies every command and architectural claim with Grep and git before writing. - Merge-safe output: Writes all guidance into AGENTS.md as the single source of truth while preserving existing content verbatim, and reduces CLAUDE.md to a pointer declaration. - Coverage reporting and safe commit: Reports read/total file counts with uncovered paths, logs discovered anomalies without fixing them, and asks the user before committing. - Use Case: After cloning a large multi-module project, ask the agent to initialize repository guidance; it dispatches read-only subagents per module, merges findings into AGENTS.md, and reports exactly what was and was not covered. ## Quick Start Initialize agent guidance for this repository by generating AGENTS.md and a CLAUDE.md pointer, verifying all claims with git before writing.

Frequently Asked Questions about repo-init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate AGENTS.md for a new repository?▼

Run the repo-init skill at the repository root. It explores the README, build files, CI config, and existing AI rules, verifies facts with git and Grep, then writes merged guidance into AGENTS.md with CLAUDE.md as a pointer.

What is the difference between AGENTS.md and CLAUDE.md?▼

AGENTS.md holds the full agent guidance as the single authoritative source. CLAUDE.md contains only the standard Claude Code prefix and a declaration pointing to AGENTS.md, with no substantive content of its own.

Does repo-init overwrite my existing AGENTS.md content?▼

No. Existing AGENTS.md content is preserved verbatim with only heading-level demotion, and new content is placed before it. Overlapping sections are merged rather than duplicated, and git diff confirms zero text changes.

Can repo-init handle large repositories with many modules?▼

Yes. It dispatches read-only exploration subagents per module or directory shard that report structured facts with file:line citations. The main agent aggregates findings and reports coverage counts plus uncovered paths.

When should I use doc-sync instead of repo-init?▼

Use doc-sync when existing documentation conflicts with actual code behavior and needs item-by-item correction. Use repo-init only for building guidance from scratch or near-scratch, since it does not perform line-by-line comparison.

Does repo-init commit changes automatically?▼

No. After writing and verifying the files, it reports the changes and asks for confirmation before committing. If there is no response, it leaves the working tree changes uncommitted and notes that confirmation is pending.

Related Skills

Explore More Software Engineering Skills→
kagentikagenti
292

onboard:link

Clone Git repositories into the Kagenti hub and verify skill discoverability.

Official
Basic
caliber-ai-orgcaliber-ai-org
1.2k

setup-caliber

Automate Caliber setup with pre-commit hooks and agent config sync.

Official
Advanced
n8n-ion8n-io
203.0k

n8n:spec-driven-development

Synchronize implementation code with architectural specification documents throughout the development lifecycle.

Official
Intermediate
filipowmfilipowm

manage-memory

Generate and update Claude Code memory files from codebase analysis.

Community
Advanced