Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

init-deep — Hierarchical AGENTS.md Knowledge Base Generator

Scans and scores your repository, then auto-generates AGENTS.md files at the root and in the subdirectories that actually deserve one.

Dev & CodingIntermediate★ 67,531⑂ 5,507AI score 7/10Last updated: Aug 9, 2026

What it does

One /init-deep command explores the whole codebase and builds a hierarchical AGENTS.md knowledge base for AI agents.

  • Discovery: fires parallel background explore agents to map structure, locate entry points, collect config-driven conventions (.eslintrc, pyproject.toml), harvest forbidden patterns (DO NOT / NEVER comments), and inspect build/CI and test setups.
  • Analysis: bash counts directory depth and file distribution while LSP (lsp_symbols, lsp_find_references) and ast-grep structural search measure symbol density, export count and reference centrality.
  • Scoring: a weighted matrix (files 3x, subdirs 2x, reference centrality 3x, …) decides placement — score >15 creates an AGENTS.md, 8–15 only for a distinct domain, <8 is skipped so the parent covers it.
  • Generate & review: the root file follows a full template (OVERVIEW, STRUCTURE, CODE MAP, CONVENTIONS, ANTI-PATTERNS; 50–150 lines), child files stay at 30–80 lines, and duplicated or generic content is stripped in a final review pass.

Flags: --create-new (read existing, then regenerate from scratch) and --max-depth=N.

Who it's for

  • Developers bootstrapping AI-agent context on a large or legacy repo with little documentation.
  • Monorepo teams that need per-package conventions surfaced to agents.
  • Maintainers refreshing stale CLAUDE.md / AGENTS.md files.

Examples

  1. Ramp up on an unfamiliar repo: run /init-deep on a freshly cloned 500-file project and get a root AGENTS.md with stack summary, directory map and a table of high-reference symbols — an instant onboarding draft.
  2. Document a monorepo: each packages/* gets scored, so only genuinely complex packages receive their own AGENTS.md and child files never repeat the parent.
  3. Reset drifted docs: /init-deep --create-new reads existing guidance for context, deletes it, and regenerates a consistent hierarchy.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the packages/omo-senpi/skills/init-deep folder from the GitHub repo code-yeongyu/oh-my-openagent into my ~/.claude/skills/code-yeongyu-init-deep/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/code-yeongyu/oh-my-openagent.git /tmp/oh-my-openagent && mkdir -p ~/.claude/skills && cp -r /tmp/oh-my-openagent/packages/omo-senpi/skills/init-deep ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and clone the repository: git clone https://github.com/code-yeongyu/oh-my-openagent.git
  2. Make sure the skills folder exists: mkdir -p ~/.claude/skills
  3. Copy the skill into place: cp -r oh-my-openagent/packages/omo-senpi/skills/init-deep ~/.claude/skills/
  4. Verify with ls ~/.claude/skills/init-deep/SKILL.md.
  5. Start Claude Code from inside the project you want documented.
  6. Type /init-deep, or simply ask: "build a hierarchical AGENTS.md knowledge base for this repo".
  7. (Recommended) Install ast-grep (brew install ast-grep) and enable LSP/MCP tooling for a far more accurate code map; without them some phases are skipped.
  8. Review the generated AGENTS.md files before committing — verify claims that the agent inferred rather than measured.
View source on GitHub ↗License: NOASSERTION