finds.dev

neural-maze/philoagents-course

When Philosophy meets AI

View on GitHub

★ 1,553PythonMITupdated Oct 2025

// the take

A six-module open course (with paired Substack articles and YouTube videos) that walks through building a LangGraph-based RAG agent which impersonates historical philosophers, wrapped in a small Phaser.js game UI. It's aimed at ML/AI and software engineers who want a full-stack example of agentic RAG rather than another notebook demo.

It actually wires together the pieces most agent tutorials skip: FastAPI + WebSockets for serving, MongoDB for short- and long-term memory, Opik for prompt versioning and evaluation, not just a LangChain chain in a notebook. The RAG grounding comes from real source material (Wikipedia + Stanford Encyclopedia of Philosophy), so the philosopher personas aren't just prompt-engineered flavor text. Tooling choices (uv, ruff, Docker Compose, langgraph.json) are current and the repo includes an actual evaluation dataset and generation script instead of hand-waving LLMOps.

This is course material, not a library or service you adopt — there's no package to install or API to depend on; you fork it and rip out what you need. It's hard-wired to Groq, MongoDB, and Opik, so swapping any one of them means touching the LangGraph nodes and the memory layer, not just a config change. The game frontend (Phaser.js) is a teaching artifact, not production UI. And like most 'N-module course' repos, momentum tends to drop once the content is finished — worth checking whether issues/PRs are still getting attention past the last recorded module.

We read the README, the directory tree and a key source file in November 2026. Nobody pays to be listed. How we judge →

// at a glance

Stars
1,553
Forks
329
Language
Python
License
MIT
Last commit
20 Oct 2025

// nearby

If philoagents-course isn't quite it.

All repos →