agentleFS
Sign inStart free
Codagent-AICodagent-AI/agent-skills/skills/handoff/SKILL.md

handoff

Skill31 starsChanged 6 months agoMITView on GitHub

Summarizes a specific aspect of the current conversation so another agent can resume. Use when the user says "handoff", "summarize for handoff", "write a handoff", or wants to capture context for another agent session. The user MUST specify what aspect to summarize in their prompt (e.g., "handoff the debugging progress", "handoff the design decisions").

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 171 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write a focused summary of a specific aspect of the current conversation so a fresh agent can pick up where this session left off.

    line 12
  2. Very specific:

    Write the summary to handoff.md.

    line 31
  3. Distinctive:

    If the user specified a path in their prompt, use it.

    line 35
  4. Distinctive:

    If a project configuration (e.g., AGENTS.md, openspec config, or similar conventions) specifies a change directory, write there.

    line 36
  5. Distinctive:

    Do not ask the user to confirm the location.

    line 39
  6. Distinctive:

    Do not invent context: Only capture what actually happened or was discussed in this session.

    line 76

“Common” is measured: kept by at least 6 in 10 of the 171 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Handoff

Write a focused summary of a specific aspect of the current conversation so a fresh agent can pick up where this session left off.

This is NOT a full conversation summary. The user's prompt tells you which aspect to capture. If the aspect is unclear, ask for clarification before proceeding.

What to capture

For the aspect the user specified, write:

  1. Objective — What the user is trying to accomplish (the goal, not the history).
  2. Current state — Where things stand right now. What's done, what's in progress, what's blocked.
  3. Key decisions — Decisions made during this session that the next agent must honor, with brief rationale.
  4. Open questions — Unresolved issues, trade-offs still being weighed, or things the user hasn't decided yet.
  5. Next steps — Concrete actions the resuming agent should take first.
  6. Relevant files — Paths to files that were created, modified, or are central to the work.

Omit any section that has nothing meaningful to say. Keep it tight — the resuming agent needs signal, not narration.

Output

Write the summary to handoff.md.

Determining the output location:

  1. If the user specified a path in their prompt, use it.
  2. If a project configuration (e.g., AGENTS.md, openspec config, or similar conventions) specifies a change directory, write there.
  3. Otherwise, determine the relevant change slug from context (the change being worked on, or a kebab-case slug from the topic) and write to ~/.agent-skills/changes/<slug>/handoff.md.

Do not ask the user to confirm the location. Just write to the resolved path.

Template

# Handoff: <aspect being handed off>

## Objective

<!-- What the user is trying to accomplish -->

## Current State

<!-- Where things stand: done, in progress, blocked -->

## Key Decisions

<!-- Decisions made this session that the next agent must honor -->
- **<decision>** — <rationale>

## Open Questions

<!-- Unresolved issues or trade-offs -->

## Next Steps

<!-- Concrete first actions for the resuming agent -->

## Relevant Files

<!-- Paths to files central to this work -->

Guardrails

  • Write from current context only. The handoff should be based on what the agent already knows from the current conversation — do not explore files, run searches, or do additional research unless the user explicitly asks for it.
  • Do not summarize the entire conversation. Focus only on the aspect the user specified.
  • Do not invent context. Only capture what actually happened or was discussed in this session.
  • Do not prescribe solutions for open questions. State the question; let the next agent and user decide.
  • Write for a cold reader. The resuming agent has zero prior context — be precise about file paths, branch names, and terminology.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free