karpathy-wiki

What problem does it solve?

Keeps a persistent, compounding markdown knowledge base so that raw research is ingested just once and every subsequent query builds on an evolving Karpathy-style wiki instead of redeveloping context from scratch.

Core Features & Use Cases

• Structured data sources: Defines raw, core knowledge, and workspace directories plus a decision tree so ingestion, queries, linting, and initialization operate consistently.
• Workflow orchestration: Coordinates summaries, concept/entity pages, index updates, and log entries to capture new material, answer complex questions, and report health without losing provenance.
• Health and automation hooks: Provides session-start drift checks and stop hooks to auto-detect new raw content, prompting ingestion when necessary while maintaining traceable logs.

Quick Start

Ask the agent to ingest new raw sources into the Karpathy Wiki so the knowledge base stays up to date.