aneja5/forge-skills

3 stars · Last commit 2026-05-21

An assembly line for AI software development. 35 skills, 11 agent personas, 29 commands. From raw idea to shipped code.

README preview

# forge-skills

> An Agentic Engineering toolkit for AI coding agents. The human writes specs and architecture. Agents implement in parallel with contracts. Review gates enforce correctness.

Structured workflows that turn a raw idea into shipped, operated, and demoed code through 7 pipeline stages, 42 skills, and 13 specialist agent personas. Every discipline-enforcing skill is pressure-tested against fresh subagents using TDD-for-skills methodology (see [Testing](#testing)).

Andrej Karpathy's Agentic Engineering concept (Feb 2026) describes exactly this model: humans write the specs, architecture, and guardrails — AI agents implement in parallel — humans review. The `.forge/` artifact chain is the implementation: `prd.md` → `architecture.md` + `contracts/` → `tasks.yaml` → code.

---

## Your AI Engineering Team

Thirteen specialist agents, available via the Task tool:

| Agent | Role |
|-------|------|
| **Architect** | System design, interface contracts, ADRs |
| **Platform Architect** | Multi-product boundaries, shared vs forked decisions, deploy topology |
| **Project Manager** | Task breakdown, dependency ordering, scope management |
| **Test Engineer** | TDD coaching, test quality review, coverage gaps |

View full repository on GitHub →