setup-meta-repo
Set up or update a meta-repo workspace containing multiple sub-repos. Creates CLAUDE.md files, .mcp.json, .claude/settings.local.json, and initializes git. Use when the user asks to "set up a meta repo", "init workspace", "setup workspace", "create workspace config", "set up meta-repo", or references setting up a multi-repo workspace.
What this skill does
# Setup Meta-Repo Workspace
You are setting up (or updating) an **organizational meta-repo** — a root directory that houses multiple independent sub-repos/projects. The meta-repo itself is NOT where feature work happens. It exists for shared configuration, Claude Code skills, and workspace organization.
## Context Awareness
This skill can be invoked at any point in conversation. Check surrounding context — the user might say "/setup-meta-repo" while already in a directory, or "set up this workspace", or reference a specific location. Use `$ARGUMENTS` if provided, otherwise use the current working directory.
## Process Overview
1. Ask clarifying questions
2. Discover sub-repos
3. Research each sub-repo
4. Create/update CLAUDE.md in each sub-repo
5. Create/update root CLAUDE.md
6. Set up .mcp.json
7. Set up .claude/settings.local.json
8. Initialize git repo on main (if not already a repo)
---
## Step 1: Ask Clarifying Questions
Use `AskUserQuestion` to gather what you need. Only ask what you can't infer from context.
Likely questions:
- **Workspace name**: What should this workspace be called? (suggest one based on directory name)
- **Sub-repos to include**: Should all subdirectories be treated as sub-repos, or only specific ones?
- **Docker Compose**: Is there a docker-compose setup? If so, which sub-repo owns it?
- **Special rules**: Any workspace-wide rules (e.g., specific sub-repo owns migrations, shared services)?
Skip questions where the answer is obvious from context or directory contents.
---
## Step 2: Discover Sub-Repos
Scan the root directory for sub-directories. **Ignore** these:
- `.claude/`
- `.git/`
- `node_modules/`
- `.vscode/`
- `__pycache__/`
- Any hidden directories (starting with `.`)
- Any known build/output directories
Each remaining directory is a potential sub-repo. Verify by checking for indicators:
- `.git/` directory inside it
- `package.json`, `requirements.txt`, `pyproject.toml`, `go.mod`, `Cargo.toml`
- `README.md`
- `Makefile`, `Dockerfile`, `docker-compose.yml`
- Source code files
List the discovered sub-repos and confirm with the user if unclear.
---
## Step 3: Research Each Sub-Repo
For each sub-repo, gather information by reading key files. Do this in parallel where possible using the Explore subagent or direct reads.
### Files to look for (read what exists):
- `package.json` — name, scripts (dev, build, test, start), dependencies, tech stack
- `requirements.txt` / `pyproject.toml` / `Pipfile` — Python dependencies
- `go.mod` / `Cargo.toml` — Go/Rust dependencies
- `README.md` — project description, setup instructions
- `Makefile` — available commands
- `Dockerfile` / `docker-compose.yml` — containerization setup
- `.env.example` / `.env.local.example` — environment variables needed
- `tsconfig.json` / `vite.config.*` / `next.config.*` — framework indicators
- `CLAUDE.md` — existing Claude instructions (if already present)
- Source directory structure (e.g., `src/`, `app/`, `lib/`, `cmd/`)
### Information to extract per sub-repo:
1. **What it is** — one-line description of purpose
2. **Technology stack** — language, framework, major dependencies
3. **How to run it** — dev server command, port, build command, test command
4. **Key directories** — where source code, tests, config live
5. **Special notes** — database ownership, migration commands, environment requirements
---
## Step 4: Create/Update CLAUDE.md in Each Sub-Repo
For each sub-repo, check if `CLAUDE.md` exists.
### If CLAUDE.md does NOT exist — Create it
Write a `CLAUDE.md` tailored to that specific repo with:
```markdown
# [Repo Name]
[One-line description of what this repo does]
## Tech Stack
- **Language**: [e.g., TypeScript, Python 3.11+]
- **Framework**: [e.g., React 18, FastAPI, Next.js]
- **Key Dependencies**: [major ones]
## Getting Started
### Install dependencies
```bash
[install command]
```
### Run development server
```bash
[dev command with port]
```
### Run tests
```bash
[test command]
```
### Build
```bash
[build command]
```
## Project Structure
```
[directory tree of key folders]
```
## Key Patterns
- [Any conventions discovered from reading the code]
## Rules
- [Repo-specific rules like "don't restart dev server", "migrations run from here", etc.]
```
### If CLAUDE.md DOES exist — Review and Update
Read the existing CLAUDE.md. Check that it includes:
- [ ] How to run the dev server (with port)
- [ ] How to run tests
- [ ] How to build
- [ ] Technology stack
- [ ] Project structure / key directories
- [ ] Any repo-specific rules
If any of these are missing or outdated based on what you discovered in Step 3, update the file. **Do not overwrite user-written content** — add missing sections, update outdated info.
---
## Step 5: Create/Update Root CLAUDE.md
Create a root `CLAUDE.md` following this structure (use the what-if example as a model):
```markdown
# [Workspace Name]
This is an **organizational meta-repo**. It exists to house shared Claude Code skills, configuration files (`.claude/`, `.mcp.json`), and workspace settings. It is **not** a repo where feature branches are created. All feature work happens in the individual sub-repos below.
## Git Rules
- **Do NOT create feature branches here.** Feature branches belong in the sub-repos.
- **Only commit here** when updating workspace-level files: `CLAUDE.md`, `.gitignore`, `.mcp.json`, `.claude/` skills, or workspace config.
- Each subdirectory is its own git repo. Always `cd` into the subdirectory before running git commands for that project.
```
[workspace-name]/ (this repo - organizational workspace)
├── .claude/ # Shared Claude Code skills & config
├── .mcp.json # MCP server configuration
├── [sub-repo-1]/ # [description] - separate git repo
├── [sub-repo-2]/ # [description] - separate git repo
└── [sub-repo-N]/ # [description] - separate git repo
```
---
## Sub-Repos
### [sub-repo-1]/
**[One-line description]**
- **Technology**: [stack summary]
- **Dev Server**: [port info if applicable]
- **Run**: `cd [sub-repo-1] && [dev command]`
- See `@[sub-repo-1]/CLAUDE.md`
### [sub-repo-2]/
[... repeat for each sub-repo ...]
---
## Key Rules
- **Git**: Separate repos per subdirectory, always `cd` first
- **Feature branches**: Create in sub-repos, never in this workspace repo
- [Any docker compose rules]
- [Any migration ownership rules]
- [Any shared service rules]
```
**Important**: The root CLAUDE.md must clearly state that **no functionality work happens in the root** — always `cd` into the specific sub-repos.
---
## Step 6: Set Up .mcp.json
Create `.mcp.json` in the root if it doesn't exist:
```json
{
"mcpServers": {}
}
```
If it already exists, leave it as-is. Tell the user they can configure MCP servers later (Linear, GitHub, Render, etc.) using the `/mcp-project-config` skill if available.
---
## Step 7: Set Up .claude/settings.local.json
Create `.claude/settings.local.json` if it doesn't exist:
```json
{
"permissions": {
"allow": []
}
}
```
If `.claude/` directory doesn't exist, create it. If settings already exist, leave them as-is.
---
## Step 8: Initialize Git Repo
If the root is NOT already a git repo:
1. Run `git init`
2. Run `git checkout -b main` (ensure we're on main)
3. Create a `.gitignore` that includes sensible defaults **and all discovered sub-repo directories**. Sub-repos are their own git repos and must NOT be tracked by the root meta-repo:
```
# Sub-repos (each is its own git repo)
backend/
frontend/
mobile/
# ^^^ Replace with actual discovered sub-repo directory names
# Standard ignores
node_modules/
.env
.env.local
*.log
.DS_Store
```
**This is critical** — without ignoring sub-repos, the root git repo would try to track them as subdirectories or submodules, which defeats the purpose of the meta-repo structure.
4. Stage the workspace files: `CLAUDE.md`, `.gitignore`, `.mcp.json`, `.claRelated in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.