Feynman: The Open Source AI Research Agent

Introduction

Feynman is an open-source AI research agent developed by companion-inc. It's designed to automate and streamline complex research tasks, acting as a powerful assistant for anyone delving into scientific papers, web resources, and codebases. With its multi-agent architecture and a suite of integrated tools, Feynman provides source-grounded outputs for various research workflows.

Installation

Feynman offers both a standalone terminal application and a "skills-only" installation for integration with other agent systems like Codex or OpenCode.

Standalone Application (macOS / Linux):

curl -fsSL https://feynman.is/install | bash

Standalone Application (Windows PowerShell):

irm https://feynman.is/install.ps1 | iex

The installer fetches the latest tagged release and includes a standalone native bundle with its own Node.js runtime. For upgrades, simply rerun the installer. Local models like LM Studio, LiteLLM, Ollama, or vLLM are supported through the setup flow.

Skills Only Installation (macOS / Linux):

If you only need the research skills without the full terminal app, you can install them for Codex, a repo-local Claude/agent, or an OpenCode project.

curl -fsSL https://feynman.is/install-skills | bash

This installs the skill library into ~/.codex/skills/feynman by default, or into specified targets like --codex, --repo, or --opencode.

Examples

Feynman simplifies complex research queries into straightforward commands, providing detailed, cited results.

What you type ? what happens:

$ feynman "what do we know about scaling laws"
- ? Searches papers and web, produces a cited research brief
$ feynman deepresearch "mechanistic interpretability"
- ? Multi-agent investigation with parallel researchers, synthesis, verification
$ feynman lit "RLHF alternatives"
- ? Literature review with consensus, disagreements, open questions
$ feynman audit 2401.12345
- ? Compares paper claims against the public codebase
$ feynman replicate "chain-of-thought improves math"
- ? Replicates experiments on local or cloud GPUs
$ feynman recipe "fine-tune a small model for math reasoning"
- ? Finds ranked, implementable ML training recipes from papers, datasets, docs, and code

Workflows:

Feynman supports various workflows, accessible via natural language or slash commands:

/deepresearch <topic>: Source-heavy multi-agent investigation
/lit <topic>: Literature review from paper search and primary sources
/review <artifact>: Simulated peer review with severity and revision plan
/audit <item>: Paper vs. codebase mismatch audit
/replicate <paper>: Replicate experiments on local or cloud GPUs
/recipe <task-or-paper>: Ranked ML training recipes with dataset, method, code, and verification status
/compare <topic>: Source comparison matrix
/draft <topic>: Paper-style draft from research findings
/autoresearch <idea>: Autonomous experiment loop
/watch <topic>: Recurring research watch
/outputs: Browse all research artifacts

Why use Feynman?

Feynman stands out as a comprehensive AI research agent due to several key features:

Multi-Agent System: It dispatches specialized agents, including a Researcher, Reviewer, Writer, and Verifier, to handle different aspects of the research process, ensuring thoroughness and accuracy.
Source-Grounded Outputs: Every output is meticulously cited, linking directly to papers, documentation, or repositories, ensuring reliability and verifiability.
Extensive Tool Integration: Feynman integrates with powerful tools like AlphaXiv for paper search and analysis, Hugging Face Hub for dataset inspection, Docker for isolated experiments, and various web search APIs, providing a rich research environment.
Local Model Support: It offers seamless integration with local large language models (LLMs) via LM Studio, LiteLLM, Ollama, or vLLM, allowing for private and efficient research.
Automated Workflows: From deep research investigations to experiment replication and ML recipe generation, Feynman automates complex tasks, significantly reducing manual effort and accelerating the research cycle.