agentleFS
Sign inStart free
desbmadesbma/agent-skills/handoff/SKILL.md

handoff

Skill1 starsChanged 16 days agoMITView on GitHub

Summarize the conversation into one self-contained handoff document per divergent path, so a fresh agent can resume any of them without loss.

191 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use decision-mapping, setup-ts-deep-modules and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • FabienDehopreSkill

    decision-mapping

    Turn a loose idea into a sequenced map of investigation tickets, then drive them to resolution one at a time.

    Also in 8 projects that use this

  • liufeng1976Skill

    setup-ts-deep-modules

    Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.

    Also in 42 projects that use this

  • AWSSkill

    to-questionnaire

    Turn a decision you can't fully answer into a questionnaire for someone else to fill in.

    Also in 93 projects that use this

  • GoogleSkill

    ask-matt

    Ask which skill or flow fits your situation. A router over the skills in this repo.

    Also in 87 projects that use this

  • RoentekSkill

    loop-me

    Grill me about specs for the workflows I want to build, within this workspace.

    Also in 51 projects that use this

  • LesaniSkill

    writing-beats

    Shape an article as a journey of beats, choose-your-own-adventure style. The user picks a starting beat from the raw material, you write only that beat, then offer options for where to pivot next, beat by beat, until the article reaches a natural end. Use when the user has raw material and wants to assemble it as a narrative rather than an argument.

    Also in 54 projects that use this

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 172 edited versions we read keep these.

  • 5

    Distinctive

    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:

    Write markdown handoff documents that let a fresh agent resume this work with no access to this conversation.

    line 10
  2. Distinctive:

    When you are done, give the user one line per document, in rank order: its absolute path, and what it carries in a few words.

    line 24
  3. Distinctive:

    Do not narrate the conversation, and never record a point the user did not settle as decided.

    line 32
  4. Distinctive:

    Do not duplicate what an artifact already holds — a spec, an issue, a commit description, a diff, code in the repository.

    line 33
  5. Distinctive:

    Do not restate the project's conventions or standing instructions: the next agent reads them from AGENTS.md, CLAUDE.md and the repository …

    line 34

“Common” is measured: kept by at least 6 in 10 of the 172 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 markdown handoff documents that let a fresh agent resume this work with no access to this conversation.

One document per path

Split the conversation into the lines of work a fresh agent could pick up on its own, and write one document per line: two unrelated bugs discussed in the conversation give two documents. Changes that belong in the same commit are one line of work and share a single document, whether they are competing fixes for one bug or closely related fixes for several. A conversation that only ever followed one line yields one document.

Files

Write them in the handoff dir, named handoff-<date>[-<rank>][-<labels>]-<slug>.md. <date> is when the set of sibling documents was first written, as YYYYMMDDHHMM, and every sibling shares it, including one added later. <labels> is the label the conversation gave each issue or fix the document covers, as the conversation wrote it, joined with -; a document whose path was never labelled has none. <slug> is a few kebab-case words naming the path: a fix for a crash on login, labelled B in the conversation and first written on 6 October 2026 at 14:30, gives handoff-202610061430-B-fix-login-crash.md.

When some paths depend on others, every document carries a <rank>, its position in an order that handles each path after those it depends on. Fixes A to D for a missing index, a login crash, a login timeout and a logout redirect, where the two login fixes share a commit that needs the index and the redirect fix builds on them, give handoff-202610061430-1-A-add-session-index.md, handoff-202610061430-2-B-C-fix-login-crash-and-timeout.md and handoff-202610061430-3-D-fix-logout-redirect.md. When all paths are independent, the documents carry no rank.

When a document for the path already exists, whatever its rank, update it in place to the state of the code and of the proposal as they stand now, but only while its path holds work left to do. Once the change it describes is implemented, do not update it; ask the user whether to delete it instead. When its rank changes, rename it, and update the pointers its siblings hold to it.

When you are done, give the user one line per document, in rank order: its absolute path, and what it carries in a few words.

Content

A document is a snapshot of where its path stands, not a log of how it got there: the problem, the state of the code, and the fix or plan the conversation landed on, which takes the bulk of it. Follow with the open points, the first step on resuming, and a few sentences at most on the options dropped along the way and why. Each document stands alone — context shared between paths is written out in full in every one of them, with a one-line pointer to its siblings, naming the ones it depends on.

Rules

  • Do not narrate the conversation, and never record a point the user did not settle as decided.
  • Do not duplicate what an artifact already holds — a spec, an issue, a commit description, a diff, code in the repository. Reference it by path or URL.
  • Do not restate the project's conventions or standing instructions: the next agent reads them from AGENTS.md, CLAUDE.md and the repository itself.

Arguments, when given, say which paths to record, or what the next session focuses on.

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.

Reviews can't be read right now.

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