---
title: Agent Pipeline
kind: open-source
role: Solo build
years: 2025 – present
status: running
summary: >-
  Agent-orchestrated development pipeline that takes a PRD to QA-ready code
  through seven discrete stages, with human checkpoints at the two decision
  points that matter.
links:
  - label: github.com/dpaola2/pipeline-skills
    url: 'https://github.com/dpaola2/pipeline-skills'
tags:
  - ai
  - claude-code
  - developer-tools
  - open-source
  - mcp
order: 1
---

**The job:** Give a coding agent enough structure that it can take a PRD to QA-ready code without me babysitting every step.

A sequence of seven distinct stages (Discovery, Architecture, Gameplan, Test Generation, Implementation, Review, QA Plan), each producing a structured artifact, with two human checkpoints: approve the architecture before tests are written, approve the gameplan before code is written. After that the agent runs.

Ships as a set of [Claude Code](https://docs.anthropic.com/en/docs/claude-code) skills (slash commands) any project can install. Repos with a conventions file and test infrastructure adopt it in minutes. The setup skill auto-detects framework, stack, and directory layout and writes a configuration block into the project's conventions file.

The principle: AI-augmented development works best when the *handoff points* are deliberate. Where humans approve, where the agent commits, where work flows between stages. Most "AI for coding" tools collapse those seams. The pipeline insists on them.

Inspired in part by Doug Engelbart's framing of system design as the place where the highest-leverage AI work actually lives. See [/writing](/writing) for related thinking, and [Work Context Protocol](/portfolio/work-context-protocol) for the persistent-memory layer this sits on top of.