Skip to content
gabrielmongefrancoPublic

About

Every great idea needs a foil. Your AI needs a second opinion. Get better results with independent AI agents that review your work and challenge each other.

Topics

Resources

Stars

1 star

Watchers

1 watching

Forks

Repository files navigation

PeerFoil™. Every great idea needs a foil.

PeerFoil

Your AI needs a second opinion. Get better results with independent AI agents that review your work and challenge each other.

AI coding with independent peer review

PeerFoil™ is an open-source workflow for AI coding and independent AI code review. It is designed to help 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.

Independent AI reviewers inspect your work, then compare findings and check each other’s contributions.

Looking for the "best ai for coding"? PeerFoil aims to bring out the best in the coding agents you use by giving them clear plans, project tests, and independent reviewers. Whether you are an experienced developer or learning as you go, the goal is better code quality with less time spent managing handoffs between AI coding tools.

The first release will connect Claude Code and Codex in a guided multi-agent workflow. You approve the important decisions while the agents handle planning, coding, testing, review, and accepted fixes. Software development comes first, with project packs planned to apply the same process to documentation, business plans, and research reports.

In development. PeerFoil has not been released yet. See Status for what is built and what comes next.

AI peer review process

Plan with independent review, produce, validate, and review. Failed checks return to production; repairs receive validation and fresh independent review before approval.

Start with reviewed architecture and a plan you approve. Produce and validate the work, then review it independently. Accepted repairs go through checks and fresh review again. PeerFoil pauses for your decision when blocked or when review limits are reached.

PeerFoil's review process gives each agent a specific job and keeps authorship separate from approval:

  • Plan before producing. Another model family reviews the architecture and plan.
  • Review AI-generated code. A fresh reviewer checks the work against your requirements and the results of project tests.
  • Check each other. Reviewers first inspect the same work independently, then compare findings. Accepted fixes receive another independent check.

An agent never approves its own work. Normal independent approval requires a qualified reviewer from a different underlying model family, not another copy of the same model. If that reviewer is unavailable, the result is labeled Reduced assurance.

The first Skills release will guide agents through these rules. Later Core releases will enforce the workflow in a local application. Peer review can improve the work, but it does not guarantee correctness or replace your judgment.

Quick Start Guide

Coming soon. PeerFoil has not been released yet.

The first version will be a set of skills for Claude Code. It will send coding and review tasks to Codex through the Codex CLI and its built-in MCP server, with no extra runtime to install. A small local command-line application written in Go will follow.

The normal workflow will be simple:

  1. Describe what you want to build or change.
  2. Answer the important questions.
  3. Review the proposed architecture and order of work.
  4. Let PeerFoil build and check one phase at a time.
  5. Step in only when PeerFoil needs a decision or approval.

Model selection, effort levels, review limits, connected tools, local models, and other controls will stay under Advanced settings.

Documentation

  • Documentation index: Every page, including setup, troubleshooting, privacy, and known limits.
  • PeerFoil Method: What PeerFoil does and how its workflow protects quality.
  • Architecture: How the planned application handles agent orchestration, coordinating models, project files, and checks.
  • Implementation Plan: What will be built first, what will come later, and how each release will be checked.
  • Phase Prompt Template: A prompt you can paste into a new chat before starting a phase or stage.
  • Decision Log: The engineering choices that later work must respect.
  • Plugin README: What the current build of the Claude Code plugin can do and how to load it for testing.

Status

Phases 1 and 2 in progress. All five stages of PeerFoil Skills 0.1 are built: the plugin, the project-file formats, setup checks, the decision interview, reviewed architecture and planning, bounded production with retained authorship and checks, plan changes, the two-family phase review with one guided repair, lessons, the three reference fixtures, and the user documentation under docs. The Core Alpha peerfoil command is implemented as a developer preview: it validates the project records, runs one software task at a time in a dedicated Git worktree, runs the declared checks itself, enforces the review and self-approval rules, and recovers at task boundaries. The live three-platform checks and real-model journeys recorded in the Phase 1 and Phase 2 plans remain open. Nothing has been released yet.

Release Target What you will be able to do
PeerFoil Skills 0.1 Day 5 Run the complete guided workflow from Claude Code
PeerFoil Core Alpha 0.2 Day 13 Use a local application that runs and records the workflow for you
PeerFoil 1.0 By Week 26 Use the complete cross-platform product, including local models and non-coding projects

The day count starts when implementation begins. Dates may change if a release does not meet its quality checks.

Example Applications

Coming soon. Planned examples include:

  • a small application built and reviewed from start to finish;
  • a documentation project with technical and editorial review;
  • a business plan with checked facts, calculations, and assumptions; and
  • a research report with source tracking and citation checks.

Common questions

Can I use Claude Code and Codex together?

That is the planned first-release setup. Claude Code will guide the workflow and use the Codex CLI's MCP server for coding and review tasks. See the Quick Start Guide and current plugin documentation for readiness and setup details.

Can PeerFoil review existing code?

The planned workflow covers changes to existing projects as well as new applications. Reviews will check the affected code, requirements, and test results. The complete review workflow is still in development.

Does PeerFoil work beyond coding?

That is part of the design. Project packs will provide instructions and checks for documentation, business plans, research reports, and other work. Each uses the same independent peer-review process, with evidence suited to the project.

Will PeerFoil support my tools and computer?

PeerFoil targets Windows, macOS, and Linux. Claude Code and Codex come first; qualified local models and broader model support are planned for later releases. It will not require a paid service other than the AI models you choose.

About the Author

PeerFoil is built by Gabriel Mongefranco, a database and software architect who has spent two decades building data platforms in healthcare and research — enterprise data warehouses, BI systems, knowledge bases, and the first architecture for mobile and wearable research data at a large research university.

Learn more at: https://gabriel.mongefranco.com

Contact

Questions, bug reports, enhancement ideas and requests are welcome as GitHub issues. Feel free to send pull requests as well!

Credits

Coming soon. Complete dependency and attribution details will be added with the first release.

This work is based in part on the following projects and libraries:

  • Codex CLI — handles coding and review tasks in the default setup and provides the MCP server that connects it to Claude Code, so PeerFoil builds no bridge of its own.
  • Git — stores project history and keeps work separated.
  • Agent Skills — provides the portable format for PeerFoil's instructions.
  • Model Context Protocol — connects approved tools and knowledge sources when a project needs them.

Versions, licenses, copied material, and required notices will be checked before release.

License

Copyright Notice

Copyright © 2026 Gabriel Mongefranco

Trademark Notice

PeerFoil™ is a trademark of Gabriel Mongefranco.

Software and Library License Notice

PeerFoil source code, tests, skills, agent definitions, project packs, templates, schemas, plugin files, configuration, and machine-consumed Markdown are licensed under the GNU General Public License, version 3 or later, unless a file says otherwise.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/gpl-3.0-standalone.html.

Documentation License Notice

Permission is granted to copy, distribute and/or modify the human-facing documentation in this repository under the terms of the GNU Free Documentation License, Version 1.3 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts, unless a file says otherwise.

See https://www.gnu.org/licenses/fdl-1.3-standalone.html.

Citation

If you find this repository or its specifications useful, please cite it.

Mongefranco, Gabriel (2026). PeerFoil™. Software and documentation. https://github.com/gabrielmongefranco/peerfoil


Copyright © 2026 Gabriel Mongefranco

About

Every great idea needs a foil. Your AI needs a second opinion. Get better results with independent AI agents that review your work and challenge each other.

Topics

Resources

Stars

1 star

Watchers

1 watching

Forks

Releases

Sponsor this project

Packages

Contributors

Languages