agentleFS
yaohuangguanyaohuangguan/agent-playbook/skills/hand-off-work/SKILL.md

hand-off-work

Skill1 starsChanged 2 days agoMITView on GitHub

Save state so that another agent or a later session can continue without guessing. Use this when you must stop before a task is finished, when context is running out, when you pass work to another agent, or when you pick up work someone else started.

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Leave the code in a known state: Make sure that it builds, or say clearly that it does not.

    line 19
  2. Distinctive:

    Write the handoff: Use the template below, and put it where the next agent will look, such as the PR description or a file the user chose.

    line 30
  3. Distinctive:

    Treat the note as a lead, not as evidence.

    line 37
  4. Distinctive:

    Check the branch, the last commits and git status.

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “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

Hand off work

The next agent knows only what you wrote down. A vague note makes them repeat your work or trust something that was never true.

When to use

  • You are stopping before the task is finished.
  • A long task reaches a natural checkpoint.
  • You are continuing from someone else's notes.

Steps

  1. Leave the code in a known state. Make sure that it builds, or say clearly that it does not. Remove temporary logging. Use a work-in-progress commit on your own branch if the user allows commits. Do not use a shared stash, because other sessions can pop it.
  2. Separate facts from guesses. Mark what you verified by running it, and what you only believe.
  3. Record what failed. Dead ends are the most valuable part of a handoff, because they stop the next agent from repeating them.
  4. Write the handoff. Use the template below, and put it where the next agent will look, such as the PR description or a file the user chose.
  5. Name the single next step. Make it specific enough to start immediately.

When you receive a handoff

Treat the note as a lead, not as evidence.

  1. Check the branch, the last commits and git status.
  2. Rerun the checks it says pass.
  3. Confirm any key claims against the code.
  4. Continue from the next step. If the note and the code disagree, trust the code, and say so.

Stop and ask when

  • The state cannot be made safe to leave, such as a half-applied data change.
  • The work needs a decision only the user can make.

Common mistakes

Tempting moveWhy it failsDo this instead
Write "almost done"Tells the reader nothingList what remains
Leave out failed attemptsThey get repeatedRecord each with its reason
State a guess as factThe next agent builds on itLabel it
Trust the previous summaryIt may be wrong or staleCheck it against the code
Leave uncommitted work without saying soIt gets lostState where the work is

Report

Goal: <one sentence>
Branch and state: <branch, last commit, uncommitted work, builds or not>

Done and verified:
- <item>: <command and result>

Done but not verified:
- <item>

Tried and failed:
- <approach>: <why it failed>

Open questions: <decisions needed>
Next step: <one specific action>

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.