plain-handoff
Skill30 starsChanged 4 months agoMITView on GitHub
Final-summary discipline for agentic sessions. Use when writing the closing message of any multi-step working session: after many tool calls, after a long build or investigation, after background or overnight work, or any time the user was not watching the work happen. Trigger on every end-of-task summary.
In this repository8 other files it gives its agents.
- delegate-and-verifySkill
- evidence-audited-analysisSkill
- finish-the-turnSkill
- lessons-ledgerSkill
- minimal-diffSkill
- outcome-first-writingSkill
- root-cause-firstSkill
- verified-doneSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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
- Distinctive: line 16
Retire your working shorthand: Every term, label, or nickname you invented mid-session either gets reintroduced in plain language or left …
- Distinctive: line 22
Write complete sentences. No arrow chains, no hyphen-stacked compounds, no fragment stacks.
- Very specific: line 24
When you mention a file, commit, flag, or identifier, give it its own plain-language clause: not "fixed via SCRAPER_TYPE in entrypoint," …
- Distinctive: line 25
If something important surfaced only mid-session (a warning, a decision you made, a tradeoff you accepted), restate it here.
- Distinctive: line 29
Before sending, verify the summary answers, in order:
- Distinctive: line 37
If you have to choose between short and clear, choose clear.
≈ “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.
Files
Plain Handoff
Your final message is the user's first look at everything you did. Write it as a re-grounding for someone who just walked in, not as the last line of a working thread only you were following.
The re-grounding rule
While working, you built up context the user never saw: file names you discovered, labels you coined, intermediate findings, dead ends. None of that vocabulary is theirs. In the final message:
- Open with the outcome. One sentence on what happened or what you found, before any detail.
- Then the one or two things you need from them, each explained as if new.
- Retire your working shorthand. Every term, label, or nickname you invented mid-session either gets reintroduced in plain language or left out. "The v2 approach" means nothing to someone who did not watch you abandon v1.
Working notes are not the deliverable
Terse shorthand between tool calls is fine; that is you thinking out loud. The final summary is a different document with a different reader:
- Write complete sentences. No arrow chains, no hyphen-stacked compounds, no fragment stacks.
- Spell out terms and acronyms on first use.
- When you mention a file, commit, flag, or identifier, give it its own plain-language clause: not "fixed via
SCRAPER_TYPEin entrypoint," but "fixed by adding theSCRAPER_TYPEenvironment variable to the startup script, which the install path had been silently dropping." - If something important surfaced only mid-session (a warning, a decision you made, a tradeoff you accepted), restate it here. Anything not in the final message effectively did not happen.
Completeness check
Before sending, verify the summary answers, in order:
- What was the goal? (One clause; the user may be returning after hours.)
- What is the outcome? (Done, partially done, blocked, and on what evidence.)
- What changed? (Files, systems, state, in plain language.)
- What do you need from the user, if anything?
- What did you deliberately not do? (Out-of-scope findings, deferred work, known limitations.)
If you have to choose between short and clear, choose clear.
Why this matters
Sessions are judged almost entirely on their final message. Hours of excellent work summarized in insider shorthand reads as confusion, and forces the user to interrogate you for the value they already paid for. The handoff is not the afterword to the work; for the reader, it is the work.
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.

