agentful

AI agent toolkit for autonomous product development with Claude Code

Quick Start • Documentation • Discord

The Swiss Army Knife of AI Agents - Works with any LLM (Claude, GLM, DeepSeek, Ollama), any tech stack, any platform. Self-hosted or cloud.

Features

• 8 specialized agents - orchestrator, architect, backend, frontend, tester, reviewer, fixer, product-analyzer
• 6 quality gates - types, tests, coverage, lint, security, dead code
• Auto-generated domain agents - learns your codebase patterns and conventions
• Self-hosted execution - run agents on your infrastructure (optional)
• Multi-platform CI/CD - GitHub Actions, GitLab, Jenkins, or any HTTP client
• Product-driven workflow - define specs, agents build features

Quick Start

# 1. Install
npx @itz4blitz/agentful init

# 2. Start Claude Code
claude

# 3. Define product spec (interactive)
/agentful-product

# 4. Start development
/agentful-start

Installation Options

# Default: All components (recommended)
npx @itz4blitz/agentful init

# Minimal: Simple scripts/CLIs
npx @itz4blitz/agentful init --preset=minimal

# Custom: Choose components
npx @itz4blitz/agentful init --agents=orchestrator,backend --skills=validation

# Web configurator
https://agentful.app/configure

Commands

Command	Description
/agentful-product	Create and analyze product specifications
/agentful-start	Start autonomous development
/agentful-status	View completion % and current work
/agentful-validate	Run quality checks
/agentful-decide	Answer blocking decisions
/agentful-analyze	Generate domain-specific agents

Self-Hosted Remote Execution

Run agents on your infrastructure:

# Deploy to Oracle Cloud free tier ($0/month)
agentful serve --auth=tailscale

# Or use SSH tunnel for local dev
ssh -L 3000:localhost:3000 your-server
agentful serve

See deployment docs for Tailscale, HMAC auth, and Oracle Cloud setup.

MCP Server

Use agentful with any MCP-compatible AI tool (Claude Code, Kiro, Aider):

{
  "mcpServers": {
    "agentful": {
      "command": "npx",
      "args": ["-y", "@itz4blitz/agentful-mcp"]
    }
  }
}

Features:

• Cross-tool compatibility (works with any MCP client)
• Launch specialized agents via MCP tools
• Access product specs and state via MCP resources
• Real-time execution status updates

See MCP Integration Guide for setup with different AI tools.

CI/CD Integration

Works with any platform via HTTP API or templates:

• GitHub Actions
• GitLab CI
• Jenkins
• HTTP API (CircleCI, Bitbucket, Travis, etc.)

Documentation

• Full docs: agentful.app
• MCP Integration: MCP Server Guide | Integration Guide
• Architecture: Agent system
• Agents: Orchestrator, Backend, Frontend, etc.
• Skills: Product tracking, Validation, etc.

Requirements

• Claude Code (code.anthropic.com)
• Node.js 22+
• Git

Tech Stack Support

Auto-detects and adapts to your stack:

• Languages: JavaScript, TypeScript, Python, Go, Rust, Java, C#, PHP, Ruby, etc.
• Frontend: React, Vue, Angular, Svelte, Next.js, Astro, SolidJS, etc.
• Backend: Express, Fastify, NestJS, Django, Flask, Spring Boot, etc.
• Databases: PostgreSQL, MySQL, MongoDB, SQLite, etc.
• Testing: Jest, Vitest, Playwright, Cypress, Pytest, JUnit, etc.

Links

• Docs: agentful.app
• GitHub: github.com/itz4blitz/agentful
• Discord: discord.gg/SMDvJXUe
• Issues: github.com/itz4blitz/agentful/issues
• NPM: npmjs.com/package/@itz4blitz/agentful

License

MIT