KEEP — Knowledge Engine for Engineering Persistence

KEEP turns a code repository into a living knowledge graph that an LLM agent can both consult and maintain. It is the answer to the question Karpathy posed for personal wikis applied to software: what if every project had a wiki the LLM kept up to date for us, that we could query reliably?

KEEP treats the repository as a cognitive system:

• code = execution layer
• /knowledge = memory layer
• agent (you) = reasoning layer

The skill is successful when a new agent or engineer can answer how and why the system evolved without re-deriving everything from code, and when changes to code automatically surface as proposed changes to knowledge.

How to trigger this skill (read this section closely)

KEEP is failing if it triggers only on code changes. Code is half the loop. The read path — consulting knowledge when answering questions — is the half that compounds value.

Strong read triggers — ALWAYS run /keep-ask first