ido4

<p align="center"> <h1 align="center">ido4</h1> <p align="center"><strong>Development Governance for AI-Augmented Teams</strong></p> <p align="center">Deterministic methodology enforcement that runs inside AI coding environments</p> </p>

<p align="center"> <a href="#see-it-in-action">See It In Action</a> &middot; <a href="#quick-start">Quick Start</a> &middot; <a href="#features">Features</a> &middot; <a href="docs/README.md">Documentation</a> &middot; <a href="#architecture">Architecture</a> </p>

<p align="center"> </p>

---

AI agents can write code. Ship features. Fix bugs. But who ensures they follow your development methodology?

ido4 is an MCP server that provides deterministic governance for AI-augmented development teams. It runs inside Claude Code (and any MCP-compatible AI environment), enforcing business rules that no agent can bypass — with full audit trails, compliance scoring, and multi-agent coordination.

Every task transition passes through 27 validation steps executed as real TypeScript code. Not LLM instructions that can be hallucinated. Not suggestions that can be ignored. Deterministic rules enforced by the AI's own tool environment.

Developer: "Start task #268"

ido4 BRE: BLOCKED — 3 validation failures:
  ✗ StatusTransition    — Blocked → In Progress is not a valid path
  ✗ DependencyValidation — dependency #267 not completed (In Progress)
  ✗ StartFromReadyForDev — task must be in Ready for Dev status

  Allowed action: unblock (the only valid path forward)

AI coding agents are everywhere — Claude Code, Copilot, Cursor, Devin — but **governance has