Skip to content

Deep Reads

Detailed analyses of key tools and papers. Each deep read follows a consistent template:

Verdict (one sentence) → What It Is → How It Works → Evaluation → Honest Assessment

Planned Deep Reads

Tools

  • [ ] Claude Code — architecture, skills system, MCP integration
  • [ ] OpenClaw — internals from contributing 4 PRs
  • [ ] LightRAG — graph extraction pipeline, AGE backend
  • [ ] Ollama — llama.cpp wrapper, model management, API

Papers

  • [ ] Attention Is All You Need (Vaswani et al., 2017) — the transformer
  • [ ] Scaling Laws for Neural Language Models (Kaplan et al., 2020) — why bigger works
  • [ ] DPO (Rafailov et al., 2023) — simpler alignment
  • [ ] Toolformer (Schick et al., 2023) — models that learn to use tools

Protocols

  • [ ] MCP specification — protocol walkthrough
  • [ ] A2A — Google's agent interop proposal

Deep reads will be added as written. Priority: tools I actively use > papers that changed my understanding > protocols I build on.