Skip to content
IFAI

Open source

Hermes Agent turns Karpathy's LLM-Wiki into a native skill

Published Reach 48

Nous Research just shipped Andrej Karpathy's LLM-Wiki pattern as a native skill inside Hermes Agent. Your agent can now automatically build persistent Markdown knowledge bases that compile, cross-reference, and lint themselves over time. Running a quick update and typing one command basically turns the agent into a tireless research librarian.

Hermes Agent turns Karpathy's LLM-Wiki into a native skill
Image: x.com

Anyone who does deep research knows the real headache isn't finding data, but maintaining it. You end up with dozens of scattered markdown files that lose their context within a week.

Nous Research just tackled this by turning Andrej Karpathy's LLM-Wiki pattern into a native skill inside Hermes Agent.

A tireless markdown librarian

The pattern allows Hermes Agent to spin up persistent markdown knowledge bases that compile, cross-reference, and lint themselves over time as new information trickles in.

Nous already dogfooded the setup on their own work. They had Hermes ingest web sources, codebase files, and Nous' own research papers to compile a complete research vault around their projects.

Using it is as straightforward as it gets. You run hermes update, punch in /llm-wiki followed by your research topic, and let the agent handle the indexing and cross-linking.

Why it matters

Most modern note-taking workflows still require massive manual labor to maintain backlinks and prune stale references. Offloading that ongoing librarian work directly into a local agent makes a ton of sense.

We think this is a huge quality-of-life upgrade for developers and researchers who need structured, self-updating documentation without being locked into proprietary SaaS platforms. That said, the real test will be whether Hermes can lint and cross-reference massive codebases without introducing hallucinated connections over extended runs.

Sources (1)

  1. [1]
    x.com(opens in a new tab)Originalx.com

Best of AI, daily