lee-to/ai-factory

AI Factory – One‑command AI‑powered development environment

What it is – AI Factory is an npm‑distributed CLI that turns a regular codebase into a fully‑featured, AI‑driven development workspace. After a single ai-factory init it installs the chosen LLM‑based coding agents (Claude, Codex, Gemini, etc.), pulls in “skills” (pre‑written prompt bundles) from the skills.sh marketplace, and configures a MCP (multi‑client‑proxy) layer that lets those agents read, edit, and commit code on your behalf.

Key capabilities

Feature	What you get
Zero‑config start	ai-factory init asks (or you pass) which agents and MCP services to use, then auto‑installs the matching skill packages.
Spec‑driven workflow	Slash commands such as /aif‑plan, /aif‑implement, /aif‑fix let the AI generate a concrete plan, execute it step‑by‑step with intermediate commits, and create documentation or CI pipelines.
Multi‑agent support	Works with a long list of commercial and open‑source coding agents (Claude Code, Cursor, Codex, Gemini CLI, etc.) and can be extended to any agent that follows the Agent Skills spec.
Community‑driven skills	Prompt bundles are published on skills.sh; you can add the whole set with npx skills add lee-to/ai-factory --skill '*' or pick individual ones.
Built‑in quality gates	After each AI action a machine‑readable aif‑gate‑result is produced, summarising lint, test, and security checks.
Extensible	Docs describe how to write extensions that add new commands, MCP integrations, or custom agents.

Typical workflow

1. ai-factory init – pick agents, install skills, configure MCP.
2. /aif‑explore – ask the AI to research a feature (e.g., OAuth login).
3. /aif‑grounded – get a verified answer before any code change.
4. /aif‑plan – AI creates a branch, analyses the repo, and outputs a step‑by‑step plan.
5. /aif‑implement – AI carries out the plan, committing after each task.
6. /aif‑fix – feed an error message; the AI patches the code and learns from the fix.
7. /aif‑ci, /aif‑docs, etc. – generate CI pipelines or project documentation automatically.

Who it’s for – Developers who want to offload repetitive scaffolding, research, and boiler‑plate coding to LLM agents while keeping the process auditable (plans, commits, quality‑gate reports). It works with any language or framework because the heavy lifting is done by the selected agent and the skill prompts, not by language‑specific tooling.

Installation – npm install -g ai-factory (or via mise). The package is versioned on npm and the repo includes CI badges showing test status.

Documentation & community – A set of markdown guides (Getting Started, Workflow, Subagents, Extensions, etc.) lives in the docs/ folder, and the project links to an official site (aif.cutcode.dev) and a skill marketplace.

License – MIT.

---

All details are taken directly from the repository’s README; no additional features are inferred.