auto-handoff

Generate deterministic 16-section handoff documents in .claude/handoffs/.

Updated Apr 29, 2026
One-click install
npx skills add https://github.com/bben15600-sys/Web-tast --skill auto-handoff-bben15600-sys
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: auto-handoff
Source: https://github.com/bben15600-sys/Web-tast/tree/main/.claude/skills/auto-handoff
Command: npx skills add https://github.com/bben15600-sys/Web-tast --skill auto-handoff-bben15600-sys

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bash, git, and includes scripts (resource) components.

What problem does it solve?

This Skill prevents lost context between sessions by generating a comprehensive handoff snapshot that lets a fresh agent continue as if the conversation never stopped. It captures the user’s voice, prior decisions, alternatives rejected, artifacts, commands, plan state, open questions, and known failure modes to reduce rework and misunderstandings.

Core Features & Use Cases

  • Periodic checkpointing: Writes an updated handoff every 20 turns to support seamless continuation.
  • Session-end capture: Produces a robust final transfer document at session end (or records a marker via the supporting hook logic).
  • Manual invocation: Generates a handoff on demand via /auto-handoff when you want an immediate snapshot.
  • Deterministic 16-section transfer format: Emits a self-contained document in a fixed section order designed for reliable resumption.

Quick Start

Ask for a save and continuity snapshot by saying: "Run the auto-handoff skill now."

Frequently Asked Questions about auto-handoff

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

FAQPage Schema
How do I preserve Claude AI session context across different conversations?

You preserve session context by generating a comprehensive 16-section handoff document that captures prior decisions, rejected alternatives, and artifacts, allowing a fresh agent to resume work without repeated questioning or lost plan state.

What is the best way to checkpoint a long coding session for later continuation?

The best way to checkpoint a coding session is to trigger periodic handoff generation every 20 turns, writing an updated snapshot of conversation continuity and plan state to a markdown file for seamless resumption.

How do I manually create a conversation continuity snapshot with git metadata?

You manually create a conversation continuity snapshot by invoking a handoff generation command, which extracts git metadata and produces a self-contained transfer document in the .claude/handoffs directory without leaking secrets.

Does session handoff work without leaking secrets or sensitive environment data?

Yes, session handoff creates deterministic transfer documents that capture workflow state and git metadata while explicitly preventing secret leakage, ensuring safe context preservation across agent transitions.

Can I use bash and git hooks to automate context preservation at session end?

Yes, you can use bash and git hooks to automate context preservation by recording session-end markers and triggering the generation of final handoff documents to capture the full conversational state for the next agent.

Why does my AI agent lose track of rejected alternatives when starting a new session?

Your AI agent loses track of rejected alternatives because standard sessions lack built-in context preservation, which requires a dedicated handoff document to explicitly capture prior decisions and known failure modes for continuity.