Ai Factory

AI Factory is a zero configuration development environment for AI coding agents. Running its init command in a new or existing project detects the stack, installs the relevant skills and configures the agents, then drives a spec driven loop of slash commands that plan, implement, verify, test, commit and improve the work. It is meant for developers and teams who want coding agents to follow a plan instead of exploring at random. Persistent specs for the roadmap, architecture and rules make each session resumable and reviewable. Planning commands turn a milestone into a branch with detailed tasks and dependencies. Implementation runs tasks one at a time with commit checkpoints, and a fix command reads the plan to understand what was intended. Verification checks each task against the plan, and a three stage QA command produces a risk annotated change summary, a scoped test plan and concrete test cases saved per branch. An evolve command folds accumulated patches back into the skills, and a distillation command turns books, documentation, folders or URLs into reusable agent skills. It works with more than 15 agents, including Claude Code, Cursor, Windsurf, Copilot, Codex, Gemini CLI, Antigravity, Roo Code, Warp and Junie, and scans external skills from the skills.sh marketplace at two levels before installing them. No paid plans are published. The tool is distributed free with its source on GitHub. The project states that the structured workflow uses about 20 percent more tokens up front on planning in exchange for fewer rewrites and fixes later.