grepai-mcp-claude_skill

Overview

This skill integrates GrepAI with Claude Code via the Model Context Protocol (MCP) to enable semantic code search and call-graph analysis inside Claude. It exposes GrepAI tools to Claude Code so the assistant can run searches, trace callers/callees, build call graphs, and check index health. Setup is a single command or a small MCP config addition.

How this skill works

The MCP server from GrepAI runs alongside Claude Code and exposes a set of compact JSON tools that Claude can call. Tools include semantic search, trace callers/callees, call-graph generation, and index status. Responses default to a compact format to reduce token usage and can be made verbose for debugging.

When to use it

• Enable semantic code search from within Claude Code for faster code understanding
• Troubleshoot or inspect call relationships using trace callers/callees tools
• Monitor index health and ensure the code index is up to date
• Integrate GrepAI into Claude workflows for multi-repository or workspace setups
• Run code searches without leaving the editor or assistant environment

Best practices

• Install GrepAI and run grepai watch to keep the index current
• Use the one-line command claude mcp add grepai -- grepai mcp-serve for quick setup
• Set cwd in the MCP config or start Claude from your project directory to target the right repo
• Prefer compact mode for routine queries to reduce token and latency costs
• Verify integration with grepai mcp-serve and grepai_index_status before heavy use

Example use cases

• Ask Claude to find all authentication-related code and return compact search hits
• Request a list of functions that call a given symbol to plan a refactor
• Generate a call graph for a critical function to assess impact before changes
• Check whether the code index is up to date after adding new files
• Run searches across multiple projects by configuring multiple MCP servers or a workspace