project-init

Generate README.md and CLAUDE.md files from detected project stack, scripts, and structure.

1|Updated Jul 2, 2026
One-click install
npx skills add https://github.com/tarabakz25/my-skills --skill project-init-tarabakz25
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: project-init
Source: https://github.com/tarabakz25/my-skills/tree/main/project-init
Command: npx skills add https://github.com/tarabakz25/my-skills --skill project-init-tarabakz25

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? New or scaffolded projects often ship code without documentation, leaving humans without a README and AI agents without a CLAUDE.md. This Skill inspects manifests, configs, directory layout, tests, and git history to generate both files grounded in what actually exists in the repo. ## Core Features & Use Cases - Automated Project Reconnaissance: Runs a detection script that parses package.json, go.mod, Cargo.toml, pyproject.toml, and config files to identify stack, scripts, entry points, and test patterns. - Dual Document Generation: Writes a human-oriented README.md (install, dev, test, build commands, structure) and a lean agent-oriented CLAUDE.md (under 100 lines, with conventions and do-not-do rules). - Safe File Handling: Detects existing files, preserves hand-written sections, and only overwrites with an explicit --force flag. - Use Case: After scaffolding a Next.js app with no docs, run the skill to detect npm scripts and the App Router layout, then produce a README with real commands and a CLAUDE.md with verified build/test instructions. ## Quick Start Ask the agent to run /project-init in your project root to generate missing README.md and CLAUDE.md files from the detected codebase.

Frequently Asked Questions about project-init

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

FAQPage Schema
How do I generate a README.md for a new project automatically?▼

Run /project-init in the project root. The skill runs a detection script that reads manifests like package.json or go.mod, identifies scripts and entry points, then writes a README with verified install, dev, test, and build commands.

How to create a CLAUDE.md file for AI coding agents?▼

Use /project-init --claude to generate only CLAUDE.md. It produces a lean file under 100 lines covering the stack, verified commands, key directories, code conventions, and project-specific do-not-do rules.

Does project-init overwrite an existing README.md?▼

No. If README.md or CLAUDE.md already exists, the skill stops and offers to enhance sections instead. It only regenerates with the --force flag, and even then it reads the existing file first to preserve hand-written content.

What project types and languages does the detection script support?▼

The script detects Node.js (package.json), TypeScript, Go (go.mod), Rust (Cargo.toml), Python (pyproject.toml), Ruby (Gemfile), and PHP (composer.json), plus frameworks like Next.js, Nuxt, Vite, and Docker via config files.

Why are some commands missing from the generated README?▼

The skill only documents commands that actually exist in manifests, Makefiles, or CI config. If a test or lint script is not defined in the repo, that section is omitted rather than guessed, so nothing fabricated appears.

When should I use codebase-onboarding instead of project-init?▼

Use codebase-onboarding when you need a deep architecture tour with data flow and request lifecycle mapping without writing files. project-init stops at generating README.md and CLAUDE.md from detected project metadata.