setting-up-a-project

Authors CLAUDE.md and AGENTS.md instruction files for coding agents in new repositories.

5|Updated Nov 9, 2025
One-click install
npx skills add https://github.com/britt/agent-skills --skill setting-up-a-project-britt
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: setting-up-a-project
Source: https://github.com/britt/agent-skills/tree/main/skills/setting-up-a-project
Command: npx skills add https://github.com/britt/agent-skills --skill setting-up-a-project-britt

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? New repositories lack the instruction files coding agents need to work reliably, causing inconsistent commands, broken workflows, and repeated rework. This Skill surveys the repo, interviews the user, and produces CLAUDE.md, AGENTS.md, TDD rules, and a verification plan before any code is written. ## Core Features & Use Cases - Repository Survey and Interview: Detects language, package manager, test tooling, and CI commands from manifests and lockfiles, then confirms findings with the user instead of interrogating them. - Multi-Agent Instruction Files: Writes CLAUDE.md for Claude Code and a self-contained AGENTS.md for Codex CLI, OpenCode, and Cursor, with no symlinks or leaked Claude-specific syntax. - Bundled Rule Files: Copies TDD.rules.md with placeholders filled, git commit rules, and Claude-specific failure patterns into the project root. - Use Case: You clone an empty repo for a TypeScript CLI. The Skill detects pnpm and Vitest, asks about purpose and branching, then commits a complete CLAUDE.md, AGENTS.md, TDD.rules.md, and VERIFICATION_PLAN.md. ## Quick Start Set up this repository for coding agents by creating CLAUDE.md and AGENTS.md with our tech stack, commands, and TDD rules.

Frequently Asked Questions about setting-up-a-project

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a CLAUDE.md file for a new project?▼

Survey the repo's manifests and CI workflows to detect the stack, then document project overview, tech stack, commands, and development practices in CLAUDE.md. This Skill walks through each section interactively and commits after every confirmed section.

What is the difference between CLAUDE.md and AGENTS.md?▼

CLAUDE.md is read by Claude Code and uses @ auto-imports plus Claude-specific tool names. AGENTS.md is read by Codex CLI, OpenCode, and Cursor, so it must be self-contained with a Required Reading list and no @ references or symlinks.

Should AGENTS.md be a symlink to CLAUDE.md?▼

No. A symlink ships Claude Code's @ imports and tool names to agents that understand neither, making them dead text. Write AGENTS.md as a real, flatter file with the same project facts written out in full.

When should I not run project setup on a repository?▼

Do not run it on an established repo that already has a mature CLAUDE.md; review and refine the existing file instead. If a thin CLAUDE.md or AGENTS.md exists, merge into it rather than overwriting.

Why does my AGENTS.md contain broken @TDD.rules.md references?▼

The @ auto-import syntax only works in Claude Code. In AGENTS.md, replace @ lines with a Required Reading list instructing the agent to open the files itself, and verify with grep that no @, TodoWrite, or .claude/ strings remain.