handoff
Skill1 starsChanged 5 months agoMITView on GitHub
Build or refresh handoff.md — a continuation doc that hands the next session (a future you, a teammate, or a run that resumes after compaction) the full picture so it keeps going instead of rebuilding context from scratch. Use when the user says "handoff", "/handoff", "write a handoff", "update the handoff", "hand off to next session", "session handoff", "save context for next time", or when wrapping up a working session that produced state worth carrying forward.
191 other projects use a version of this.
Adopted by
Most recently taken up by
- xiaoxiaosrm/dsh-mattpocock-skillstheir own version, first seen Nov 8, 2026
- ethosagent/ethostheir own version, first seen Nov 8, 2026
- dsifry/metaswarmtheir own version, first seen Nov 8, 2026
- glebis/claude-skillstheir own version, first seen Nov 8, 2026
- domengabrovsek/agent-configtheir own version, first seen Nov 8, 2026
- domengabrovsek/claudetheir own version, first seen Nov 8, 2026
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.
FabienDehopreSkilldecision-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
liufeng1976Skillsetup-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
AWSSkillto-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
GoogleSkillask-matt
Ask which skill or flow fits your situation. A router over the skills in this repo.
Also in 87 projects that use this
RoentekSkillloop-me
Grill me about specs for the workflows I want to build, within this workspace.
Also in 51 projects that use this
LesaniSkillwriting-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
Other versions191 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 191
- mattpocock/skills★ 309ksame name
- QuivrHQ/quivr★ 40ksame name
- The-Vibe-Company/quivr★ 40ksame name
- GreyDGL/PentestGPT★ 16ksame name
- ykdojo/claude-code-tips★ 10ksame name
- ArcReel/ArcReel★ 6.3ksame name
- vinvcn/mattpocock-skills-zh-CN★ 5.5ksame name
- withcoral/coral★ 4.9ksame name
- epicenter-md/epicenter★ 4.8ksame name
- EpicenterHQ/epicenter★ 4.8ksame name
181 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository24 other files it gives its agents.
- authorship-validatorSkill
- auth-setupSkill
- code-reviewSkill
- design-reviewSkill
- design-systemSkill
- devex-reviewSkill
- diagram-siteSkill
- doc-syncSkill
- file-followup-issuesSkill
- idea-reviewSkill
- investigate-bugSkill
- plan-reviewSkill
- pr-commentSkill
- pre-merge-reviewSkill
- qa-fixSkill
- qa-reportSkill
- qaSkill
- scope-reviewSkill
- semantic-compressorSkill
- shipSkill
- sql-performanceSkill
- team-retroSkill
- ui-polishSkill
- weekly-planSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.
0
Common
Most of the 172 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 31
Look for handoff.md at the repo root (ls handoff.md):
- Very specific: line 36
If $ARGUMENTS points at a focus (e.g.
- Distinctive: line 109
Don't pad for the appearance of thoroughness.
“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.
Files
Session Handoff Skill
Produce or update handoff.md at the project root: a compact, skim-friendly picture of the current situation so the next session is productive in seconds instead of minutes — without re-deriving what's already known or re-breaking what already works.
The richest source of material is this session's own recollection of what it did. Write the handoff right here in the main conversation — do not hand the authoring off to a subagent, which carries none of the conversation's context. Tools exist only to pull durable repo facts that confirm or correct that recollection.
A handoff is neither a changelog nor a memory dump. It exists to answer four things for whoever picks up next: Where do we stand? What just changed? What's still open? What's the first move? Anything that doesn't serve one of those gets cut.
Gather context first
This skill ships standalone, so it can't assume your project's conventions, house style, or preferences the way it could for its author. Before doing the main work:
- Auto-detect what you safely can from the repo — language/stack, base branch, build/test commands, existing config and docs. Never ask for something you can read for yourself.
- Ask, don't assume, for the rest. Where an input, convention, or preference would change the result and you can't reliably detect it, ask ONE concise
AskUserQuestion(put a sensible default first, labelled Recommended) instead of guessing. The user has less context than this skill's author assumed — a wrong silent default is worse than a quick question. Don't ask about things you can detect, and don't ask more than you need.
For this skill, confirm up front (only the items you can't already detect):
- Output path & filename — default is
handoff.mdat the repo root; confirm if this project keeps continuation/session docs somewhere else or under a different name. - Commit or leave uncommitted — auto-detect from
.gitignore;handoff.mdis usually per-session scratch left untracked, but confirm before staging if the project tracks such docs. - Focus / scope — if
$ARGUMENTSdidn't name one, which in-flight thread the next session most needs carried forward (vs. a general "where we are" pass over everything). - Where open work and durable notes live — which review / TODO / issue / notes files to point at; the house
PLAN_REVIEW_*/REVIEWED.md/TODOS.mdnames may not exist here, so point at whatever this project actually uses instead of inventing them.
Step 0 — create vs update
Look for handoff.md at the repo root (ls handoff.md):
- Found → UPDATE it in place. Do NOT overwrite wholesale. The expensive part is the operational context — re-verify commands, infra coordinates, traps, "do NOT do X" warnings — and preserving whatever is still true is the entire point. Then: refresh the dated header; swap "What shipped" for this session's work; re-confirm or explicitly mark stale every runtime value the previous handoff stated; carry over still-open follow-ups and drop the resolved ones; rewrite "Next session". If this session disproved something the old handoff claimed, call it out (
~~old~~ → corrected: …) rather than quietly replacing it. - Missing → CREATE one from the template below.
If $ARGUMENTS points at a focus (e.g. handoff focus on the migration), tilt the doc toward that thread while still answering all four questions.
Step 1 — gather the state (facts, not impressions)
Replay the current conversation for what genuinely happened this session, then check it against the repo:
git branch --show-current
git log --oneline -15
git status --short
git log --oneline @{upstream}..HEAD 2>/dev/null # unpushed local commits
gh pr list --state open --limit 10 2>/dev/null # open PRs
gh pr list --state merged --limit 8 2>/dev/null # recently merged
ls *REVIEW*.md REVIEWED.md TODOS.md handoff.md 2>/dev/null # review artifacts + existing todosDraw on these, highest priority first:
- The conversation itself — what got decided, built, fixed, deferred, and why. Only the live session holds this, which is exactly why this skill can't run as a subagent.
- Open-work trackers — whatever this project actually uses: a
TODOS.md, review files (PLAN_REVIEW_*.md,SCOPE_REVIEW_*.md,REVIEWED.mdif present), GitHub issues, or aNOTES.md. Detect what exists rather than assuming the house review-file names; cite them by name and don't restate what they already contain. - Durable project notes — any long-lived memory or notes file the project keeps for facts already recorded. Link to them; never copy their contents across.
- Deploy / runtime surface — if the project has a
deploy.sh, aninfra/directory, or any live/deployed component (the project's CLAUDE.md or README usually says), capture the Current state section. For a library or CLI with no runtime, skip it and note build/test status instead.
Confirm every cited fact before it goes in — PR numbers, commit SHAs, file paths, host IPs, env-flag values. A handoff that points the next session at the wrong place, branch, or box is worse than none at all. For any runtime value you can't re-confirm right now, record it as (last verified <date>, re-check before trusting) rather than asserting it as current.
Step 2 — write the handoff
Front-load whatever prevents wasted effort and broken things. Trim the rest. Aim for ~1–2 screens; a handoff nobody reads is dead weight. Include only the sections that apply:
# Session Handoff
**Date:** <absolute date or range, e.g. 2026-06-24>
**Branch:** <branch> — <clean / open feature branch / everything merged>
**State:** <one line: is it deployed/live/healthy? or for non-runtime projects: tests green, build OK, mid-refactor?>
---
## TL;DR
<3–6 sentences. The paragraph to read if you read nothing else: where things stand,
what this session changed, and the single most important thing to know going in.>
## Current state — verify before trusting ← only if there's a live/deployed/stateful component
- **Infra:** <host, key, dir, the env flags that matter>.
- **Last verified (<timestamp>):** <the values an operator would trust and act on>.
- **Re-verify:** `<the exact command to re-check the above>`
(State is stale the moment it's written — this line is mandatory for anything live.)
## What shipped this session
| PR / commit | Content |
|---|---|
| #NN (merged) | <one line> |
| `<sha>` | <one line> |
## Open follow-ups
- <unresolved item> → tracked in `TODOS.md #N` / issue #N / `<review file>`. (Point, don't re-explain.)
## In-flight / not landed
- <branch or research track>: <what's done, what's left, how to resume it>.
## Gotchas / notes
- <trap that bit you this session, environment quirk, or an explicit "do NOT do X — it Y">.
## Next session — start here
1. **<the single highest-priority next action, named concretely>** — <why / first command>.
2. <next>…Style
- Skimmable: short bullets, bold the load-bearing words, tables for PR lists.
- Absolute dates only (turn "yesterday" / "next week" into actual dates).
- Provenance refs (PR numbers, commit SHAs, ticket IDs) are welcome here —
handoff.mdis throwaway per-session scratch, so naming them costs nothing. The code those refs point at, though, should still read clearly on its own. - Name the one thing to tackle first, in plain terms. "Various cleanup remains" helps no one.
- Don't pad for the appearance of thoroughness. A small session earns a small handoff.
Output
Write handoff.md to the repo root (or wherever this project keeps continuation/session docs, if it has a different convention — confirm rather than guess). Print a 2-line confirmation: what it now covers plus the named first step for next time. Do not commit it unless asked — handoff.md is usually left uncommitted (or gitignored) as per-session scratch; check .gitignore and the repo's convention before staging.
When this fires on its own
It pairs well with context pressure: when a session has run long and is nearing compaction, offer to write the handoff before the context is lost. It complements any durable project-notes update — those hold the long-lived facts, while the handoff captures the volatile "where we are right now" the next session needs. Don't auto-write without a prompt; offer first.
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
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.

