project-handoff
Skill0 starsChanged 3 months agoMITView on GitHub
Create or update a concise, evidence-backed HANDOFF.md in the correct user-confirmed project directory before context is cleared or work is handed to another agent. Use only when explicitly invoked as $project-handoff; never invoke implicitly.
8 other projects use a version of this.
Adopted by
Most recently taken up by
- zhouey314-cloud/ai-agent-skillstheir own version, first seen Oct 20, 2026
- duoduoler-ops/Table-skillstheir own version, first seen Oct 4, 2026
- NinjaSln-labs/agent-skillstheir own version, first seen Oct 2, 2026
- LURENLEO/project-handofftheir own version, first seen Sep 29, 2026
- ChenneyZhuang/project-handofftheir own version, first seen Sep 23, 2026
- tranhaianh20571-droid/project-handoff-skilltheir own version, first seen Sep 19, 2026
Other versions8 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 8
- duoduoler-ops/Table-skills★ 437same name
- HeathTeng/project-handoff★ 2same name
- kim-unho/project-handoff-skill★ 2same name
- LURENLEO/project-handoff★ 1same name
- NinjaSln-labs/agent-skills★ 1same name
- ChenneyZhuang/project-handoff★ 0same name
- tranhaianh20571-droid/project-handoff-skill★ 0same name
- zhouey314-cloud/ai-agent-skills★ 0same name
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.
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. 0 are common, 11 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
11
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Create a current snapshot that lets a new agent continue the work without the conversation.
- Distinctive: line 14
Prefer an explicit directory supplied with the invocation.
- Very specific: line 16
Treat the current working directory, Codex workspace root, version-control root, and project markers such as README.md, AGENTS.md, …
- Very specific: line 18
If the destination is missing or ambiguous, show the likely candidates and ask one concise question: HANDOFF.md 要写到哪个目录? Do not guess.
- Very specific: line 20
If multiple independent projects were changed, prefer a separate HANDOFF.md for each and confirm their destinations before writing.
- Distinctive: line 24
After resolving each target directory:
- Distinctive: line 27
If version control exists, inspect its status and relevant changes.
- Distinctive: line 28
Use existing test or validation evidence.
- Very specific: line 29
Read an existing HANDOFF.md before updating it.
- Distinctive: line 31
Do not rely only on conversation memory when filesystem evidence is available.
- Very specific: line 35
Create or update <confirmed-target>/HANDOFF.md as a concise current-state document with these sections:
- Distinctive: line 64
Use absolute dates, concise prose, and paths instead of transcript-style narration.
- Distinctive: line 65
Never include credentials, tokens, private personal information, environment-file contents, or large raw logs.
- Distinctive: line 66
Keep the file as the latest handoff snapshot rather than an append-only history.
- Distinctive: line 75
every required section is present.
- Distinctive: line 78
no sensitive information was copied into the document.
“Common” is measured: kept by at least 6 in 10 of the 4 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
Project Handoff
Create a current snapshot that lets a new agent continue the work without the conversation. Do not clear the conversation yourself.
1. Resolve the destination
Determine the handoff target before writing:
- Prefer an explicit directory supplied with the invocation. Normalize it and verify that it exists.
- Otherwise, infer candidates from directories and files actually discussed, read, created, or edited in this session.
- Treat the current working directory, Codex workspace root, version-control root, and project markers such as
README.md,AGENTS.md,package.json, orpyproject.tomlonly as signals. None is authoritative by itself, and a project need not use Git. - Proceed without asking only when one candidate is clearly supported by the session evidence.
- If the destination is missing or ambiguous, show the likely candidates and ask one concise question:
HANDOFF.md 要写到哪个目录?Do not guess.
If multiple independent projects were changed, prefer a separate HANDOFF.md for each and confirm their destinations before writing.
2. Inspect the actual state
After resolving each target directory:
- Review the relevant session work and inspect the target's structure, instructions, key documentation, implementation files, and outputs.
- If version control exists, inspect its status and relevant changes. If it does not, continue without treating that as an error.
- Use existing test or validation evidence. Run only lightweight, safe checks when useful; do not install dependencies or make unrelated project changes.
- Read an existing
HANDOFF.mdbefore updating it. Keep still-valid information, replace stale status, and move newly completed work out of the remaining-work section.
Do not rely only on conversation memory when filesystem evidence is available.
3. Write HANDOFF.md
Create or update <confirmed-target>/HANDOFF.md as a concise current-state document with these sections:
# Project Handoff
> Updated: YYYY-MM-DD
> Target: /absolute/confirmed/path
## Current state and structure
## Completed cases
## Decisions and conventions
## Remaining work and exact next step
## Pitfalls and dead ends
## Validation, risks, and open questionsApply these rules:
- Give completed cases concrete outcomes and project-relative file or directory paths.
- Record decisions, technical choices, style or naming conventions, and their reasons. Include evidence locations when useful.
- Make remaining work actionable: priority, exact starting file or command, blockers, and the condition for completion.
- Record failed approaches and why they failed so the next agent does not repeat them.
- State tests or checks run and their results. Explicitly say what was not verified.
- Distinguish observed facts from session claims and inferences. Mark uncertainty as
待确认; never fabricate completion, decisions, or test results. - Use absolute dates, concise prose, and paths instead of transcript-style narration.
- Never include credentials, tokens, private personal information, environment-file contents, or large raw logs.
- Keep the file as the latest handoff snapshot rather than an append-only history.
Modify only the intended HANDOFF.md unless the user explicitly requests other changes.
4. Verify and report
Re-read the finished file and confirm that:
- it is in the confirmed directory;
- every required section is present;
- the next action is specific enough for a new agent to begin immediately;
- claims, paths, dates, and validation status agree with the available evidence;
- no sensitive information was copied into the document.
Report the written path, a brief summary, and any unresolved uncertainty. Stop there so the user can review the handoff before manually running /clear.
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.

