yeachan-heo/oh-my-claudecode
Overview
This skill performs deep codebase initialization by generating hierarchical AGENTS.md documentation across every directory. It creates AI-readable notes that explain purpose, key files, subdirectories, dependencies, and agent-specific instructions. The output preserves manual annotations and keeps parent references valid to form a navigable hierarchy.
How this skill works
The skill recursively maps the project tree, excluding common build and vendor folders, then produces AGENTS.md files level by level so parents exist before children. For each directory it reads files, infers purpose and relationships, generates structured sections (Purpose, Key Files, Subdirectories, For AI Agents, Dependencies), and writes or merges content while preserving manual notes. It validates parent links, detects orphans, and can skip empty or generated-only directories according to configurable rules.
When to use it
- Onboarding new engineers or agents to a large or unfamiliar codebase
- Preparing a repository for multi-agent automation and task delegation
- Maintaining up-to-date, AI-friendly documentation across many nested modules
- Before running automated refactors or large-scale changes to ensure context
- As part of CI to detect structural drift and missing documentation
Best practices
- Run generation from project root and ensure parent directories are accessible
- Exclude node_modules, .git, build artifacts, and other large vendor folders
- Preserve manual notes below the <!-- MANUAL --> marker to keep human context
- Process same-level directories in parallel but generate parents before children
- Review and adjust brief file descriptions rather than relying on generic boilerplate
Example use cases
- Generate a root AGENTS.md that lists top-level manifests, scripts, and main subtrees
- Create nested AGENTS.md under src/ that explains core modules, components, and utilities
- Automatically update AGENTS.md after adding or removing directories and preserve manual sections
- Skip empty feature folders while creating minimal AGENTS.md for container directories
- Validate all AGENTS.md parent references and remove or fix orphaned documentation files
FAQ
It recognizes a <!-- MANUAL --> marker and preserves any content below that line while only updating auto-generated sections and timestamps.
What directories are skipped?
By default it excludes node_modules, .git, dist, build, __pycache__, .venv, coverage, .next, .nuxt and can be customized for additional folders.
35 skills
This skill autonomously transforms a brief idea into requirements, design, code, tests, and validation, delivering working software without manual
This skill generates comprehensive, hierarchical AGENTS.md documentation across the codebase to guide AI agents through structure and dependencies.
This skill helps you extract reusable decision-making insights from conversations to improve Claude's problem-solving approach.
This skill orchestrates consensus planning by coordinating Planner, Architect, and Critic to deliver validated plans with user-approved execution.
This skill orchestrates parallel task execution across multiple agents, routing to appropriate model tiers to save time and tokens.
This skill routes prompts to local Claude, Codex, or Gemini CLIs and saves the results as reusable artifacts for later review.
This skill guides unified setup, diagnostics, and MCP configuration for Claude Code orchestration, streamlining installation and environment readiness.
This skill conducts Socratic deep interviews to uncover assumptions and ensure crystal-clear specs before autonomous execution.
This skill launches multiple tmux CLI workers (claude, codex, or gemini) to execute tasks in parallel and return structured results.
This skill guides you through configuring Telegram, Discord, or Slack notifications via natural language, saving time and ensuring reliable alerts.
This skill orchestrates Claude, Codex, and Gemini to decompose, parallelize tasks, and synthesize a cohesive backend and frontend solution.
This skill orchestrates parallel external context searches by routing facets to document-specialist agents and synthesizing results for actionable insights.
This skill orchestrates multiple Claude Code agents to decompose tasks, assign roles, and coordinate subtasks across a team for faster delivery.
This skill orchestrates parallel scientist agents to perform comprehensive research workflows with optional auto mode for autonomous execution.
This skill helps writers maintain persistent character, scene, world, and theme memory across sessions, boosting consistency and collaborative workflows.
This skill helps you automate isolated development environments using git worktrees and tmux sessions for parallel task work.
This skill helps configure and optimize the HUD display in Claude Code to reflect status, presets, and elements for your workflow.
This skill automates QA cycling, running tests, builds, lint, and typechecks until the quality goal is met, with autonomous diagnosis and fixes.
This skill helps you query Codex via a local CLI and capture reusable artifacts for multi-agent workflows.
This skill displays the agent flow trace timeline and summary, helping you diagnose interactions among hooks, keywords, skills, agents, and tools.
This skill initializes a structured PRD for Ralph loop execution, guiding goal-driven iterations and aligning teams with acceptance criteria and milestones.
This skill analyzes your OMC usage to deliver personalized recommendations for optimizing workflows and feature usage.
This skill saves important context to a persistent notepad, ensuring notes survive conversation compaction and are easily retrievable.
This skill configures OpenClaw to wake external automations and AI agents on Claude hook events.
This skill lets you consult Gemini locally and capture reusable advisory artifacts for iterative AI-assisted coding workflows.
This skill fixes build and TypeScript errors with minimal changes to get the project compiling without refactoring.
This skill accelerates complex, multi-module tasks by orchestrating parallel subtasks with exclusive file ownership to unlock up to 5x speed.
This skill orchestrates sequential and branching AI agent pipelines, passing structured context between stages to automate complex code tasks.
This skill helps you manage isolated dev environments with git worktrees and tmux by delegating to the project session manager.
This skill routes swarm invocations to the Team skill, enabling coordinated multi-agent task execution with familiar interfaces.
This skill performs thorough codebase search and synthesis across files, imports, and usage patterns to identify core implementations and connections.
This skill guides you through configuring Slack incoming webhooks and channels to deliver alerts when sessions end or tasks complete.
This skill configures Telegram notifications for OMC by guiding setup and saving a usable config for alerts on session events.
This skill guides you through configuring Discord notifications with webhook or bot, writing settings for automated alerts.
This skill helps you manage git history with atomic commits, rebasing, and branch hygiene to streamline collaboration.
