Never lose session context again. agent-handoff persists decisions, technical details, open questions, and next steps across agent resets, compactions, and new sessions.
agent-handoff creates structured memory files your agent loads at the start of each new session.
Save full session context as a structured Markdown + JSON file in one command. File list, decisions, next steps, open questions all captured.
Load any prior checkpoint at session start. The agent reads the structured file and is immediately oriented without any re-explaining.
Search all prior handoffs for a file path, decision, or keyword. Never wonder "did I implement that already?" in seconds.
Hand off between different AI agents: from Copilot to Cursor, or from a subagent back to the main agent with full fidelity.
List all handoffs in chronological order with one-line summaries. Understand the arc of a long project at a glance.
All handoff files stay on your machine. No cloud sync, no telemetry. The format is plain Markdown, readable without any tooling.
Python 3.8+. Zero runtime dependencies.