handoff
Skill0 starsChanged 51 days agoMITView on GitHub
Condenses the current session into a short handoff note a fresh session can resume from, pointing at existing artifacts instead of copying them, separating verified state from assumptions, and recording decisions with their reasons so they aren't re-argued. Run manually with /handoff, optionally naming what the next session will focus on.
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 repository9 other files it gives its agents.
- adversarySkill
- claritySkill
- contextSkill
- dev-councilSkill
- impactSkill
- reflectSkill
- scopeSkill
- technical-writingSkill
- verifySkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 5 are distinctive, and 2 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 21
If the user named a focus for the next session, shape the note around it and leave out threads that don't bear on it.
- Distinctive: line 31
Every claim about state is one of two things, and the note says which:
- Distinctive: line 44
Not a list of possibilities.
- Very specific: line 58
Same machine, later: Write it to the system's temp directory ($TMPDIR, falling back to /tmp, or %TEMP% on Windows) with a descriptive, …
- Distinctive: line 60
The user named a location: Use it.
- Distinctive: line 62
If the host can start a new session seeded with a prompt, offer that as an option.
- Very specific: line 102
scope: if a scope statement exists, point to it rather than restating the boundaries.
“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
Handoff
A long session gets more expensive every turn, because every turn resends everything before it. Starting fresh is the cheapest fix, but only if the fresh session doesn't have to rediscover what this one already learned. This skill writes the note that makes starting fresh safe.
The note is not a transcript summary. It's the minimum the next session needs to act correctly: where things stand, what was decided and why, what's still open, and what to check before trusting any of it.
When to use it
- The session is long and the next piece of work doesn't need most of what's in it.
- You're switching to a different task and will come back to this one later.
- Work is moving to another agent, another machine, or another person.
- Context is close to being compacted, and you'd rather choose what survives than let a summary choose for you.
If the user named a focus for the next session, shape the note around it and leave out threads that don't bear on it. Mention them in one line so they aren't lost silently.
Writing the note
1. Point, don't copy
Anything that already exists somewhere durable gets a pointer, not a restatement: commits, branches, PRs, issues, specs, scope statements, design docs, files in the repo. The next session can open a path. Copying an artifact into the note doubles its cost and creates a second version that can go stale. The note holds only what exists nowhere else: the reasoning, the dead ends, the current mental model.
2. Separate what's verified from what's assumed
Every claim about state is one of two things, and the note says which:
- Verified. Checked in this session, with how: "tests pass, ran
<command>", "PR open, checked on the host". - Assumed / not checked. Believed but not confirmed, or confirmed long enough ago that it may have drifted.
A fresh session trusts whatever it's handed. An unverified claim written as fact becomes the next session's first bug.
3. Record decisions with their reasons, and the options rejected
This is what gets lost most often and costs the most to rebuild. For each decision that shaped the work: what was chosen, why, and what was considered and rejected, in a line each. Without the rejected options, the next session rediscovers them, finds them plausible, and reopens a question that was already settled.
4. End with one next action
Not a list of possibilities. The single most useful next step, concrete enough to start on immediately. Other open threads go in their own section.
5. Keep it small
The note should be a small fraction of the session it replaces. Roughly a screen is a good target for most work. If it's growing well past that, something is being copied that should be pointed to. Cut detail before cutting threads: a thread the next session doesn't know about is worse than one it knows about briefly.
6. Redact
Strip credentials, tokens, keys, personal data, and anything the user marked private. The note may be pasted somewhere with a wider audience than this session had.
Where the note goes
Choose based on where the next session will run, and say where the note went:
- Same machine, later. Write it to the system's temp directory (
$TMPDIR, falling back to/tmp, or%TEMP%on Windows) with a descriptive, timestamped filename. It stays out of the repo, so nothing gets committed by accident. - Somewhere else, or an ephemeral environment (a cloud container, CI, a remote session that will be reclaimed). A temp file won't survive. Give the note in the reply so the user can carry it, or, if the user agrees, save it where the next session can reach it. Don't commit a handoff note to a shared branch without asking.
- The user named a location. Use it.
If the host can start a new session seeded with a prompt, offer that as an option. The note becomes the prompt, and redaction matters even more.
The note
HANDOFF: [what this work is, one line]
Written: [date] Next focus: [what the next session is for, if given]
Resume instructions for the next session:
Read this note, run the checks under "Verify on pickup", and report any
mismatch before acting on anything below.
State:
- [branch / PR / key files, as pointers] (verified: how | assumed)
Decisions:
- [decision]: [why]. Rejected: [option], [why not]
Dead ends:
- [approach tried]: [why it didn't work]
Open threads:
- [thread, one line each]
Verify on pickup:
- [command or check that confirms the state above is still true,
e.g. branch exists and is clean, PR still open, tests still pass]
Suggested skills:
- [only skills installed in this session, and why each applies]
Next action:
[one concrete step]Drop any section that has nothing in it, except "Verify on pickup" and "Next action", which are always present. A handoff with nothing to verify is claiming the world didn't change between sessions.
Interaction with other skills
contextkeeps a single session lean. This is for when a session has done its job and the work should continue in a fresh one.scope: if a scope statement exists, point to it rather than restating the boundaries.verify: an evidence block written earlier in the session is the best source for the "verified" lines. Point to it.reflectcaptures lessons worth keeping beyond this piece of work. A handoff is about continuing this work; if something in it deserves to outlive the task, route it throughreflectinstead.
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.

