FounderOS Agent Catalog

This catalog maps the workflow to the agent instructions currently in the repository. “Human owner” means a founder or project team member makes and approves the decision; an AI agent may help prepare the artifact.

Workflow stage	Current owner / agent	Main handoff	Coverage
Idea and market research	Founder; Product Discovery Agent organizes evidence	Evidence, assumptions, opportunity assessment → discovery	No dedicated research agent; starter template available
Product discovery	Product Discovery Agent	Discovery document → PRD	Agent available
Requirements	PRD Agent	Approved PRD → architecture and planning	Agent and starter template available
Solution architecture	Solution Architecture Agent	System boundaries and decisions → specialist design	Agent available
Domain model	Domain Modeling Agent	Business concepts and rules → database and API	Agent available
Database design	Database Design Agent	Schema and migration strategy → implementation	Agent available
API specification	API Specification Agent	Contracts → backend, frontend, and tests	Agent available
Security architecture	Security Architecture Agent	Threat model and controls → design, implementation, tests, release	Agent available
UI/UX specification	Founder / product team; Frontend Module Implementation Agent consumes the approved design	Flows, pages, states, accessibility → frontend implementation	Starter template available; no dedicated specification agent
AI behavior and prompt design	Founder / product owner approves behavior; Solution Architecture Agent records system constraints; AI Module Implementation Agent implements	AI contract, data rules, eval criteria → AI implementation	Starter template available; no dedicated AI design agent
Implementation planning	Implementation Planning Agent	Sequenced module plan → implementation	Agent available
Backend implementation	Backend Module Implementation Agent	Increment → Testing Agent	Agent available
Frontend implementation	Frontend Module Implementation Agent	Increment → Testing Agent	Agent available; depends on approved UI/UX decisions
AI implementation	AI Module Implementation Agent	AI module and evaluation evidence → Testing Agent	Agent available; a longer engineering guide also exists at agents/engineering/ai-module-implementation-agent.md
Testing	Testing Agent	Test evidence and defects → Code Review Agent	Agent available
Code review	Code Review Agent	Review decision and findings → Release Agent	Agent available
Release and deployment	Release Agent	Release evidence and learnings → team	Agent available; human retains production approval
Growth and continuous improvement	Founder / product team, with Release Agent handing off observations	Outcomes and learnings → next discovery cycle	Starter learning template available; no dedicated agent

Agent document conventions

The long-form engineering and architecture guides generally include dependencies, a detailed output artifact, quality gates, a definition of done, and a prompt. Some newer concise guides use a shorter structure. Treat the role's explicit inputs and quality gates as the handoff contract, and surface missing inputs before work begins.

Duplicate AI implementation guides

Two AI implementation guides currently exist: the concise guide in agents/ai/ and the detailed guide in agents/engineering/. They describe the same role at different levels of detail. Use the concise guide as the quick entry point and the engineering guide for its full workflow and output structure until the repository consolidates them.