#llm-agents

Local memory + orchestration layer for custom LLM agents: SCOPE-style tactical/strategic memory over execution traces, Qdrant-backed, polyglot subprocess tools, agent-as-tool composition.

Local-first checkpoints and crystals for long-running AI coding agent sessions.

Agent Native Format (ANF): a token-minimal view format that translates live system state into far fewer tokens for AI agents. Ships a Go encoder, a Kubernetes translator, and a governed execution runtime. v0.1 PoC.

An agentic software factory that closes the loop

Sovereign single-binary knowledge mesh for coding agents (MCP): index, search and serve your notes + code as context. Fair-code (no reselling).

Read live tmux panes as deterministic JSON for LLM coding agents — capture, diff, and Claude/Codex/Copilot status. Single Go binary, no network egress.