Summary

This wiki is a working synthesis layer for fast-moving technical knowledge about agentic systems. It is designed to accumulate conclusions, comparisons, patterns, contradictions, and decision triggers across repeated research sessions.

Key points

• It is intentionally markdown-native and simple.
• It is optimized for compounding understanding, not just storing files.
• It complements OpenClaw memory and corp KB rather than replacing them.
• Its highest value comes from repeated questions across evolving tools, models, and runtime patterns.

Operational relevance

Use this repo when you need:

• a stable map of the local-model and agentic tooling landscape
• durable comparisons across frameworks and runtimes
• explicit notes on tradeoffs, gaps, and shifting beliefs
• reusable answers to recurring technical questions

Open questions

• Which Linear AI Dev issues should seed the first serious ingest wave?
• Which external sources should be mirrored locally first?
• What lint cadence is enough to control drift without creating ceremony?

Sources

• docs/FIT_ASSESSMENT.md
• wiki/sources/karpathy-llm-wiki-gist.md