LLM notes
My journey with LLMs, agents, tools and more.
Agent workflows and skills
- Cache control - how to cut agentic sessions costs using explicit cache control.
- Matt Pocock skills - grill-me, grill-with-docs, tdd, diagnose, etc.
- Code doctrine - The Reusable Version of Kamil Chmielewski AI Coding Constitution
- Obra Superpowers - A complete software development workflow for your coding agents, built on top of a set of composable “skills” and some initial instructions […]
- The Open Agent Skills Ecosystem
- Agent Skills
- Home Assistant Skills
- Fabric - AI-powered prompts and patterns for everyday tasks
Agent harnesses and orchestration
- Pi Dev - Minimal terminal coding harness, that adapts to a developer’s specific workflow. Unlike more rigid agents, it provides[…]
- Oh My OpenCode - Multi-agent orchestration layer for the OpenCode platform that uniquely transforms single-stream AI coding into a parallelized team[…]
- Open Agents Control - AI agent framework built specifically for the OpenCode ecosystem that prioritizes human-in-the-loop control and production-g[…]
- Plannotator - Interactive plan annotation and colaborative sharing.
Context, memory, and knowledge
- Opencode Memory - Long-term context retention across sessions using local vector database
- Context7 MCP
- How We Built an AI Second Brain for 60K Knowledge Workers
- Local-first AI journal
MCP servers and integrations
Token usage and cost optimization
- Coding (Agent) CLI Usage Analysis - Analyze coding (agent) CLI token usage and costs from local data
- RTK - CLI proxy that reduces LLM token consumption
- Caveman - Talk like a caveman and save the tokens
- llmfit - One command to find what runs on your hardware