agentleFS
nathanonnnathanonn/agent-skills/skills/handoff-doc/SKILL.md

handoff-doc

Skill22 starsChanged 5 months agoMITView on GitHub

Write or update a session handoff doc that lets a future session (Claude or human) resume work without re-reading the whole transcript. Use this skill whenever the user asks to "write a handoff", "create a handoff doc", "update the handoff", "write a doc so the next session can continue", "session handoff note", "hand this off", or points at a path and says to capture where things stand. The user only needs to give the destination file path — the content is distilled from the current session. Do NOT use for general project documentation, READMEs, API docs, code comments, or changelogs.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    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 concise handoff doc that captures the current session's work so a future session can pick up cold.

    line 8
  2. Distinctive:

    Look at the path the user gave and the folder it lives in.

    line 21
  3. Distinctive:

    Sanity-check the filename against today's date: If the date in the filename looks wrong (e.g.

    line 27
  4. Distinctive:

    If the target doesn't exist (or is an empty placeholder), create it.

    line 34
  5. Very specific:

    If the target already exists with content, don't assume.

    line 35
  6. Distinctive:

    Read back over what happened this session and pull out what the next session actually needs.

    line 47
  7. Distinctive:

    Use this skeleton as a starting point, not a cage.

    line 53
  8. Distinctive:

    Be honest about gaps: If something wasn't tested, say so.

    line 84
  9. Distinctive:

    Link sequels: If this continues a prior handoff, name it and point to it so the chain is followable.

    line 90

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Handoff Doc

Write a concise handoff doc that captures the current session's work so a future session can pick up cold. The whole point is leverage: instead of the next session re-reading a long transcript, it reads one tight doc and knows what was done, where things live, what's verified, and what's still open.

The user gives you a destination path. You supply the content by distilling what actually happened in this session. Keep it concise — a handoff is a launch pad, not a transcript.

Workflow

1. Resolve the target path

Look at the path the user gave and the folder it lives in.

  • List sibling files in the target directory. Many handoff folders use a YYYYMMDD_NN_short-slug.md convention (date, then a two-digit sequence). If the siblings follow a pattern, match it — continue the numbering and use today's date.
  • Sanity-check the filename against today's date. If the date in the filename looks wrong (e.g. a year or month that doesn't match today, or it's out of order with its siblings), it's probably a typo. Surface it with AskUserQuestion and offer the corrected name — never silently rename.

2. Decide create vs. update

  • If the target doesn't exist (or is an empty placeholder), create it.
  • If the target already exists with content, don't assume. Use AskUserQuestion to ask how to proceed, with these options:

    • Overwrite in place (replace the contents).
    • Append a new dated section to the same file.
    • Sequel — write a new file at the next _NN_ in the series, referencing the prior one.

    This is a genuine fork with no safe default — losing a handoff someone wanted kept is worse than asking one question.

3. Distill the session into the doc

Read back over what happened this session and pull out what the next session actually needs. Default audience is a future Claude session, and default scope is as-built state plus open follow-ups — these fit most cases, so just proceed with them unless the conversation makes the audience or scope genuinely ambiguous (then ask).

Use this skeleton as a starting point, not a cage. Keep the sections that carry weight for this session; drop ones that don't apply; add one if the work needs it. A small infra change won't need a "Git state" section; a big refactor might need an "Architecture decisions" one.

# Handoff: <subject> (<date>)

Audience: <who this is for>. <If a sequel: one line pointing to the prior doc.>

## What happened this session
<The work done, in the order it matters. Decisions made and why.>

## Where things live / what landed
<Files created/changed, key paths, where the important code is.>

## Verification done
<What was tested/checked and the result — and honestly, what was NOT verified.>

## Git state
<Branch, commit hash, pushed or not, how to ship.>

## Open follow-ups
<Numbered. What's left, including anything offered-but-not-confirmed. Carry
forward still-open items from a prior handoff.>

4. Keep it concise and honest

  • Concise by default. Favor short sections and tight bullets over prose. The reader wants orientation fast, then the details when they dig in.
  • Be honest about gaps. If something wasn't tested, say so. A handoff that hides an untested area sets the next session up to trust something it shouldn't.
  • Make paths and commands copy-pasteable. The next session will act on them.
  • Convert relative time to absolute. "Yesterday" is meaningless to a session that runs next week — write the date.
  • Link sequels. If this continues a prior handoff, name it and point to it so the chain is followable.

When to ask vs. just write

This skill leans autonomous — the user's whole ask is "give a path, get a doc," so don't interrogate them. Reserve AskUserQuestion for the cases where a wrong guess is costly:

  • An existing file at the target (overwrite / append / sequel — always ask).
  • A filename whose date or sequence looks like a typo (offer the fix).
  • Genuinely ambiguous audience or scope that the conversation doesn't settle.

Everything else — section choices, what to include, phrasing — is yours to decide from the session. Write the doc, then tell the user where it landed and give a one-line summary of what it covers.

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.