Engineering infrastructure for AI agents: prompt-to-loop pattern libraries, persistent codebase memory, agent fleet orchestration, terminal multiplexers, and skills that turn books into callable tool calls.
Upgrades raw prompts into running loop systems. Ships 7 production-grade patterns and 3 CLIs.
Compresses a codebase into a Markdown knowledge graph. CLI semantic search across the whole repo.
Indexes a codebase into a persistent knowledge graph with sub-millisecond queries. Claims 99% token reduction vs. raw RAG.
Agent fleet orchestrator. Parallel multi-agent execution with desktop and mobile clients.
Terminal multiplexer for agents, written in Rust. tmux for AI agents.
Obsidian plugin that embeds Claude Code directly into your notes.
Compiles a technical book into a callable Skill. Reports 15.6x token savings vs. RAG over the same content.
Microsoft's AI terminal: an agent panel that reads shell output and supports multiple agent CLIs.