Handoff
Compact the current conversation into a handoff document so the next agent or session can continue without reading the history. Use at the end of any phase, or when the user is wrapping up and wants a clean starting point for the next session.From its SKILL.md
npx -y skills add RubenGlez/harness --skill handoffAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.4 KB, 521 tokens by cl100k_base, as published. Nobody here has run it
Handoff
What to produce
A single markdown document that gives a fresh agent everything it needs to pick up exactly where this session left off — without reading this conversation.
Reference, don't duplicate. Do not re-paste content that already exists in .harness/ files or other artifacts — link to them instead.
Step 1: Gather what exists
Identify:
- Every
.harness/file written or updated this session (path + one-line summary of what changed) - Every code file created or modified (path + what changed)
- Any public root files updated (README.md, DESIGN.md, CHANGELOG.md)
- Any outstanding issues or blockers
- Any decisions made in conversation that are NOT yet captured in
.harness/adr/
Step 2: Write the handoff document
Save to the OS temp directory — never to the repo:
- Resolve path:
$TMPDIR→/tmp→%TEMP%(Windows) - Filename:
handoff-[YYYY-MM-DD-HHmm].md
# Handoff — [YYYY-MM-DD HH:mm]
## Context
One paragraph: what this project is, where it sits in the development lifecycle, and what this session accomplished.
## What was done this session
- [artifact or action] — [one-line summary]
## Current state
- **Phase**: ideation / product / engineering / implementation / qa / docs
- **Features done**: [list, or "none yet"]
- **Features in progress**: [list, or "none"]
- **Blockers**: [list, or "none"]
## Key decisions made in conversation
Only decisions NOT already in .harness/adr/. Omit ephemeral reasoning ("chose X because it was faster today") — only capture things that would surprise a future agent.
## Artifacts to read first
In priority order for the next agent:
1. [path] — [why it matters]
2. ...
## Suggested next skill
/[skill-name] — [one sentence on what it will do and why now]
## Notes for next agent
Anything surprising, any implicit constraints, or any context that doesn't emerge from reading the artifacts alone.
Omit any section that has nothing to say. Redact secrets, API keys, and PII before writing.
Step 3: Tell the user
Output the full absolute path to the handoff file. If the user wants to start a new session, they can paste the file contents as the first message.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.
Gives 0 of the 12 instructions most agent orchestration skills give in 521 tokens
Counted across 848 of the 1,300 authors here whose files we hold, read 2026-09-06
- Dispatch one agent per independent problem domainin 56 of 848, across 42 files
- Run full test suite after integrationin 55 of 848, across 42 files
- Verify fixes do not conflictin 40 of 848, across 32 files
- Review each summary when agents returnin 40 of 848, across 31 files
- Write a handoff document summarising the current conversationin 30 of 848, across 25 files
- Reference existing artifacts by path or URLin 26 of 848, across 24 files
- Give each agent a specific scopein 19 of 848, across 10 files
- Give each agent a clear goalin 19 of 848, across 10 files
- Include a suggested skills section in the documentin 18 of 848, across 16 files
- Tailor the doc to the user argumentsin 18 of 848, across 15 files
- Issue all subagent dispatches in the same responsein 17 of 848, across 11 files
- Use git worktrees for isolationin 17 of 848, across 8 files
Said here and by no other author read
- Gather existing session files and changes
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.