“summarization”

Opinionated agent harness on LangGraph with sub-agents, filesystem access, context management, persistent memory and approval of tool calls.

Lightweight Python framework and starting template for agents that let the model choose its own approach to each task.