pm-llm-wiki

Stop re-deriving your product knowledge. Start compounding it.

This is Karpathy's LLM Wiki pattern adapted for product management: an AI agent (Claude Code, Codex, or any coding agent) maintains a wiki of your product's problems, decisions, assumptions, and open questions — fed by your customer interviews, feedback, analytics, competitor intel, and meetings. You curate what goes in and make the calls; the agent does all the filing, cross-referencing, and maintenance.

The payoff: your PRDs, prioritizations, stakeholder briefs, and "why did we decide X?" answers are generated from accumulated, source-cited evidence — not rewritten from memory and vibes every time.

To the best of our knowledge, this is the first open-source instantiation of the LLM Wiki pattern purpose-built for product management — Karpathy published the pattern as an abstract idea file in April 2026, a rich ecosystem grew around it for engineers and researchers, but no one had built the PM version: the schema, the entity model, the rituals, and a worked example. That matters because PM work is arguably the best fit for this pattern anywhere:

• Deliverables become queries. A PRD or stakeholder brief that used to take days of

re-gathering context is generated in minutes from evidence that's already synthesized, cross-referenced, and cited.

• Every claim becomes defensible. "Users need X" links to who said it, when, and in what