I will set up claude code for your repo with claude md, skills, hooks and mcp

Claude Code is powerful, but on a real codebase it guesses, breaks conventions and skips tests unless the repo is set up for it. I set up your repository so the agent works reliably.

What I set up:

• CLAUDE.md with your architecture, commands and rules
• Path-scoped rules for frontend, backend and tests
• Custom skills (slash commands) for common tasks: add an API endpoint, write tests, review a change
• Hooks: auto-format, lint, block risky commands
• Permission allowlist so you approve less
• One verify script every change must pass (lint, build, test)
• MCP servers for your tools (GitHub, database, browser)

Why me: I built the Claude Code setup my team uses on two production products, a clinical diagnostics platform and an EdTech SaaS: an 11-skill ticket pipeline (spec, plan, implement, verify, review), agent boundary rules and hooks. My own projects use the same pattern with a single verify gate.

Message me your stack and repo size.