rusnino

AI Software Factory — AI skill for Claude Code

AI community

Self-hosted, open-source AI Software Factory: a Governance Controller that orchestrates heterogeneous coding agents (OpenCode, Claude Code, Codex, Aider) with durable human approval gates, a determini.

How to install AI Software Factory

This entry records only its repository, not the path inside it, so there is no exact command to give. Open rusnino/ai-software-factory and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Software Factory does

Self-hosted, open-source AI Software Factory: a Governance Controller that orchestrates heterogeneous coding agents (OpenCode, Claude Code, Codex, Aider) with durable human approval gates, a deterministic state machine, and an append-only audit log. Plane CE is the human UI — the Controller stays authoritative.

Alternatives in AI

  • Better Harness — Review this project's AI coding workflow with Better Harness and generate a durable report 2.1k ★
  • Agentrq — AgentRQ: Human-in-loop realtime conversational task manager for AI Agents 1.1k ★
  • Claw Empire — Command Your AI Agent Empire from the CEO Desk — A local-first AI agent office simulator that orchestrates CLI 1.1k ★

README

AI Software Factory

Self-hosted, open-source, free AI Software Factory for orchestrating heterogeneous coding agents.

Goal

A human submits a high-level idea, reviews decomposed tasks, approves execution, and lets heterogeneous coding agents do the work in isolated workspaces — with explicit human gates and deterministic governance.

Core Principles

  • Self-hosted — runs on infrastructure we control.
  • Open-source + free — no license fees, minimal TCO.
  • Human-in-the-loop — human approval before execution and before final merge.
  • Governance authority — custom Controller owns approvals, state machine, policy, audit.
  • Plane as projection — Plane CE is the human UI, not the workflow authority.
  • Heterogeneous agents — OpenCode (primary), Claude Code, Codex CLI, Aider, and more.

Architecture Layers

Human
  ↓
Plane CE (human project layer / projection)
  ↓ (webhooks)
Governance Controller (authoritative)
  ↓
macro-agent (execution orchestration)
  ↓
OpenCode / Claude Code / Codex / Aider
  ↓
Git + CI + Reviewer + Human Review → Done

Repository Layout

Path Purpose
specs/ Design specifications (SPEC-01..SPEC-10)
decisions/ Architecture Decision Records
requirements/ Requirements traceability matrix
docs/ Supporting docs, diagrams, deferred-risk notes
reviews/ Review findings (REVIEW-NNN-*.md); the cross-review gap ledger GAPS.md is frozen/superseded — gap tracking moved to GitHub Issues, see AGENTS.md § Review Findings and Gap Tracking
src/governance_controller/ Phase 1 Governance Controller (FastAPI/SQLModel/PostgreSQL); its own tests/ live inside this subproject, not at repo root. See the live gh issue list --label phase-1 / --label phase-2 for the current gap ledger, not reviews/GAPS.md.
tmp/ Generated archives, diagrams, scratch files (ignored by git)

Documentation

Start with:

  1. specs/SPEC-01-vision.md
  2. `specs/SPEC-0