**Feynman — CLI Tool for AI-powered research flows**

Spending hours on research and still couldn’t find any good results or come up empty? With the speed of light the AI advancements are rapidly evolving and today I discovered a new tool called Feynman. ”Feynman ships as a standalone runtime bundle for macOS, Linux, and Windows, and as a package-manager install for environments where Node.js is already installed. The recommended approach is the one-line installer, which downloads a prebuilt native bundle with zero external runtime dependencies” It’s an easy-to-use tool. It has a few tools already built in there such as Alphaxiv, Web Search, Session Search, Preview. It has prebuilt agents too which are easy to use. Researcher — Which focuses on information gathering

Reviewer — Evaluates the academic papers, documents and research papers with the sternness of an academic peer review.

Writer — The writer agent will help you in structuring your research findings in the proper formatted document.

Verifier — Helps in validating and fact checking.

The UI of it will look like below:

You can play with different models/LLMs but you need to have an API key for any LLM to utilise it. It has to be exported via .env file or via export command.

Also, If in case you want to update the model api key you can always go to .feynman/agents/settings.json here you can update your default model details.

It will look like something below:

{
 “packages”: [
 “npm:@companion-ai/alpha-hub”,
 “npm:pi-subagents”,
 “npm:pi-btw”,
 “npm:pi-docparser”,
 “npm:pi-web-access”,
 “npm:pi-markdown-preview”,
 “npm:@walterra/pi-charts”,
 “npm:pi-mermaid”,
 “npm:@aliou/pi-processes”,
 “npm:pi-zotero”,
 “npm:@kaiserlich-dev/pi-session-search”,
 “npm:pi-schedule-prompt”,
 “npm:@samfp/pi-memory”,
 “npm:@tmustier/pi-ralph-wiggum”
 ],
 “quietStartup”: true,
 “collapseChangelog”: true,
 “defaultThinkingLevel”: “medium”,
 “editorPaddingX”: 1,
 “theme”: “feynman”,
 “defaultProvider”: “anthropic”,
 “defaultModel”: “claude-opus-4–6”,
 “lastChangelogVersion”: “0.62.0”,
 “images”: {
 “blockImages”: true
 }
}

To update the api key for the specific model again in the agent directory you have auth.json file there you can update the api key. Below is an example file of auth.json:

{
  "your model name": {
    "type": "api_key",
    "key": "your api key"
  }
}

You can also select the model as per your choice using command ‘/models’.

I tried out with the claude model. It worked really well. I have experimented with the /deepresearch workflow…..on transformer attention mechanisms it pulled out the papers and the github repos as well + It has generated the 2-page brief summary with citations within 3 minutes.

It has a lot more features to explore. Multiple workflows available to use as well.

Here are a few:

Deep Research : Multiple agents working in parallel to search for the academic papers, web sources, code repositories and then after that it combines all findings into a structured brief with the citations for the same.

Code audit: This one I personally find a cool tool that would quite help researchers since it compares what any research paper claims and then it checks the actual codebase that would be available publicly and flags if there would be any discrepancy found.

There are more workflows in it. I would suggest go checkout here and play with it on your terminal. PhD scholars, your job of researching has been made easier by these AI advancements.

References:

https://www.feynman.is/docs/getting-started/installation

https://github.com/getcompanion-ai/feynman