The idea
PeerFoil is an open-source workflow for AI coding with independent AI code review. It helps solo developers use different AI model families to plan, create, test, and review their projects together. Agents check your work and each other's contributions to catch mistakes a single model might miss.
Two reviewers from different model families look at the same work, then compare what they found.
How the review works
Work starts with a reviewed architecture and a plan you approve. The agents do the work and check it. Then a new reviewer from another model family checks it against your requirements and the results of the project tests. When a fix is accepted, it goes through the checks and a new review again. PeerFoil stops and asks you to decide when it is blocked or when it reaches its review limits. The agent that wrote a piece of work never approves it.
The workflow. Each fix goes back through the checks and a new review, and no agent approves its own work.
Status
PeerFoil has not been released yet. The first release will connect Claude Code and Codex in a guided multi-agent workflow. It starts with software development. Project packs for documentation, business plans, and research reports are planned. The repository documents the workflow, the architecture, and the implementation plan.
