agentup-cli helps teams bootstrap consistent AI coding-agent instructions in any repository. Instead of manually writing agent rules every time, run one command and generate production-ready context files.

GitHub About SEO (Recommended)

Suggested repository description:

CLI to generate AGENTS.md and AI coding agent scaffolding for Claude Code, Cursor, Codex, and Gemini.

Suggested GitHub topics:

agentup, agentup-cli, agents-md, ai-agent, ai-coding-agent, cli, claude-code, cursor, codex, gemini, prompt-engineering, context-engineering

Why AgentUp

• Generate AGENTS.md automatically for AI coding assistants.
• Scaffold provider-specific configs for Claude Code and Cursor.
• Keep agent roles consistent (plan, review, test, code) across repos.
• Detect project stack (language, framework, database, package manager) to produce relevant defaults.
• Reduce onboarding time for new contributors and AI-assisted workflows.

What AgentUp Generates

Always generated:

• AGENTS.md
• .agentup.json

Generated when Claude Code is selected:

• CLAUDE.md
• .claude/settings.json
• .claude/rules/*
• .claude/commands/*
• .claude/skills/*
• .claude/agents/*

Generated when Cursor is selected:

• .cursor/README.md
• .cursor/rules/*
• .cursor/commands/*
• .cursor/skills/*
• .cursor/agents/*
• .cursor/docs/architecture.md

Install

Global install:

npm install -g agentup-cli

Run without global install:

npx agentup-cli@latest init

Quick Start

cd your-project
agentup-cli init

The interactive CLI asks for:

• provider targets: Claude, Codex, Cursor, Gemini, Antigravity
• primary IDE
• context source: auto-detect or manual
• content mode: ai or template
• roles: plan, review, test, code
• overwrite mode: ask, skip, replace

If overwrite mode is ask, AgentUp asks before changing each existing file.

AI Mode (Gemini)

ai mode uses Gemini review to improve generated context.

cp .env.example .env
# set GEMINI_API_KEY in .env

agentup-cli automatically loads .env and .env.local from the selected project root. If GEMINI_API_KEY is missing, it safely falls back to template mode.

CLI Commands

agentup-cli init        # Interactive scaffolding
agentup-cli __selftest  # Built-in self-tests
agentup-cli --version   # Show installed version
agentup-cli --help      # Show command help

Check latest npm version:

agentup-cli --version
npm view agentup-cli@latest version

Example Output

your-project/
├── AGENTS.md
├── .agentup.json
├── CLAUDE.md                # if Claude selected
├── .claude/                 # if Claude selected
└── .cursor/                 # if Cursor selected

Local Development

npm install
npm run dev -- init
npm run build
npm run selftest
node dist/index.js init

Automated npm Release

GitHub Actions workflow: .github/workflows/publish.yml

Release by pushing a version tag:

VERSION=$(npm run -s version:show)
git tag "v$VERSION"
git push origin "v$VERSION"

Workflow checks:

• npm ci
• npm run build
• tag version matches package.json version

Required GitHub secret:

• NPM_TOKEN

Version bump helpers:

npm run release:patch
npm run release:minor
npm run release:major

FAQ

What is AgentUp CLI?

AgentUp is an AI agent setup CLI that generates repository instructions and provider-specific scaffolding for coding assistants.

Can I use AgentUp for Codex, Claude Code, and Cursor?

Yes. AgentUp supports multi-provider generation in one run.

Does AgentUp work in existing repositories?

Yes. Use overwrite mode ask or skip to safely adopt it in active projects.

License

MIT