handoff

Record durable learnings and continuation notes in a MEMORY.md index.

20|5|Updated Nov 30, 2025
One-click install
npx skills add https://github.com/tale-project/tale --skill handoff-tale-project
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: handoff
Source: https://github.com/tale-project/tale/tree/main/.claude/skills/handoff
Command: npx skills add https://github.com/tale-project/tale --skill handoff-tale-project

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Preserve durable learnings and concise continuation context across context boundaries and agent handoffs, preventing loss of decisions and non-obvious facts.

Core Features & Use Cases

  • Persist durable learnings to memory when a task spans sessions or requires handing off to another agent.
  • Compress context for quick re-orientation, including a concise continuation note and pointers to relevant files.
  • Use cases include long-running experiments, multi-agent coordination, and knowledge retention across iterations.

Quick Start

Record a key decision and a concise continuation note for the next agent.

Frequently Asked Questions about handoff

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I preserve context and decisions when handing off tasks between agents?

To preserve context during agent handoffs, record explicit memory notes and a concise continuation context. This approach prevents the loss of durable learnings and non-obvious facts across sessions.

Why does my agent lose key decisions when the context limit is reached?

Agents lose key decisions near context limits because durable learnings are not explicitly recorded. Compressing context into lightweight continuation notes and a centralized index prevents this loss.

What is the best way to manage memory across multi-session workflows?

Managing memory across multi-session workflows requires a centralized index and explicit memory recording. This setup ensures durable learnings and continuation context are preserved for future iterations.

Can I use a single memory index for long-running experiments and multi-agent coordination?

Yes, you can use a centralized memory index for long-running experiments and multi-agent coordination. It enforces explicit memory recording and provides pointers to relevant files for quick re-orientation.

How to compress context for quick re-orientation of a new agent?

Compress context for a new agent by creating a concise continuation note with pointers to relevant files. This lightweight summary enables quick re-orientation without reloading the full history.