Feynman
Overview
Open-source CLI research agent that reads papers, searches the web, drafts research outputs, plans experiments, and cites claims.
Feynman is an open-source AI research agent from Companion for scientific and technical research workflows. It runs from a REPL or shell as commands such as deep research, literature review, critique, reproducibility audit, replication planning, comparison, drafting, and bounded autoresearch. The agent gathers evidence from literature databases, web pages, PDFs, Hugging Face Hub, alphaXiv, and local documents, then links research claims back to sources.
AI Agent Store research
What the evidence says about Feynman
Feynman is an open-source research agent focused on scientific-literature workflows, combining paper, web, code, PDF, and local-document evidence gathering with cited research drafts and review-oriented workflows.
Verified October 10, 2026
Verified capabilities
- Scientific research automation
The official page says Feynman reads papers, searches the web, writes research drafts, plans experiments, and cites research claims.[1]
- Research workflows
It exposes workflows including deep research, literature review, internal review, reproducibility audit, replication planning, ML recipe generation, source comparison, drafting, and bounded autoresearch.[1]
- Plan-first deep research
The /deepresearch workflow is described as a plan-first investigation across papers, web, and code, with parallel researchers when the topic needs them.[1]
- Evidence source coverage
The official page lists literature sources including Semantic Scholar, OpenAlex, arXiv, PubMed, Europe PMC, bioRxiv/medRxiv, and Crossref, plus alphaXiv, Hugging Face Hub, web research, PDFs, and local documents.[1]
- Cited outputs
Feynman states that research claims link to their sources and that its draft workflow can produce paper-style drafts with inline citations from findings.[1]
- Command-line and REPL operation
Workflows can run as slash commands in the REPL or as feynman workflow commands from the shell.[1]
- Agent-oriented architecture
The GitHub search result describes the project as running on stock Pi with four research agents: researcher, verifier, reviewer, and writer.[2]
Where it fits best
- Researchers who need literature reviews from primary sources with consensus mapping.[1]
- Technical teams investigating topics across papers, web sources, and code using a plan-first deep-research workflow.[1]
- ML and reproducibility teams that want paper-to-code mismatch audits, replication plans, or ranked ML training recipes grounded in papers, datasets, docs, and code.[1]
- Writers of research drafts who want polished paper-style drafts with inline citations from findings.[1]
Buying and deployment notes
Feynman is presented as an open-source installable research agent; the supplied context does not show paid tiers, a price point, or a separate pricing page.[1], [2]
Platforms: CLI, REPL, local shell[1]
Deployment: Open-source local installation, Standalone bundle via curl or PowerShell with bundled Node.js, npm installation using local Node.js[1]
Important considerations
- The supplied official context presents Feynman as a command-line and REPL tool with curl, PowerShell, and npm installation paths; it does not show a hosted web app access model.[1]
- The official install note distinguishes standalone bundles with their own Node.js from npm installation that uses the local Node.js, so setup requirements depend on the chosen installation route.[1]
- No paid pricing page, paid tiers, or commercial support terms are shown in the supplied context, so organizations should verify licensing and support needs before adoption.[1], [2]
Sources and research method (3)
We record only claims tied to public sources checked by our team or listing workflow. Counts above are derived directly from this profile, not a subjective rating.
- Feynman — The open source AI research agentOfficial site · checked 2026-10-10
- Companion-Inc/feynman: The open source AI research agentGitHub · checked 2026-10-10
- Release Notes — Feynman DocsChangelog · checked 2026-10-10
Autonomy level
76/100
Our subjective rating of how independently the product carries out its documented tasks, considering the human setup, input, approval and review it needs. This is not a measured success rate.
Comparisons
Custom Comparisons
Some of the use cases of Feynman:
- Literature reviews from primary sources
- Plan-first deep research across papers, web, and code
- Research draft generation with inline citations
- Paper-to-code reproducibility audits
- Replication planning and experiment planning
- Comparing sources for agreements and conflicts
Loading Community Opinions...
Pricing model:
Code access:
Popularity level: %
Tags:
Feynman Video:
Turn this profile into a trust signal
Show prospects that Feynman has a public place where they can check product details, pricing, ratings, and reviews.
Build confidence
Give buyers a third-party profile to explore.
Reduce hesitation
Put validation beside your strongest CTA.
Earn discovery
Every badge links prospects to your listing.
Choose your style
Preview it, then copy the complete embed code.
Shows buyers where to validate your product, pricing, and reputation.
Plain HTML. No signup, script, or maintenance required.
Describe the job. Get an AI worker you can actually message.
We create the setup, keep it running after your laptop closes, and save its memory. Test in the browser, then add Telegram, WhatsApp, or Slack.
Plans start at $29/month. Cancel anytime.
Hosted agent
OpenClaw or Hermes