agentsclimarketplace

Handoff

Skill keli-wen/skills/skills/handoff

Agent skills I use day to day (lowband / handoff / multi-agent-driven / ctx-* family)

Install
npx -y skills add keli-wen/skills --skill handoff

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

3 things to look at

  • 20 days oldThe repository was created 20 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 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.

What its author says it does

Copied from the file, not written here

Generate a minimal-sufficient handoff when another agent or session will take ownership of a task or continue unfinished work. Use for /handoff, "把任务交给下一个 agent", "生成 handoff", "让另一个 session 继续", session transfer, task ownership transfer, or restarting in a fresh context when the current session has degraded (a handoff to your future self). Reuses durable artifacts by reference instead of copying them, and redacts secrets.

SKILL.md

4.0 KB, 741 tokens by cl100k_base, as published. Nobody here has run it

handoff

Compress the current task into a prompt the next agent can start from directly. Issues, context docs, PRs, commits, and diffs are the sources of truth; the handoff only provides entry points, current state, and the necessary not-yet-persisted delta.

The same compression works in three directions: transferring ownership to another agent or session; disciplining spawn contracts when a dispatcher delegates slices (see multi-agent-driven); and restarting yourself — when the current context has degraded (noise from failed explorations polluting decisions), a handoff aimed at your own fresh session beats continuing in the noise. Compaction is just a handoff to your future self.

Language: mirror the user's language.

Core rules

  • Reuse durable artifacts first; never copy an existing spec, issue, or checklist into a fourth source of truth.
  • Return the handoff in the conversation by default; write a file only when the user asks.
  • No fixed length. The standard is: the next agent can start working, and no fact is duplicated.
  • Do not create or update issues, write to external systems, or send messages without explicit consent.
  • A handoff request does not imply authorization to commit, push, open PRs, or change remote state.

Workflow

  1. Define the transfer: the next agent's goal, what is in and out of scope, and whether the task is not-yet-started or in progress.
  2. Discover durable artifacts (read-only): tracking issue, linked PR, authoritative context/ADR docs, current branch/worktree, commits, diff, latest verification results. Record each artifact's path or URL and status; do not restate its content.
  3. Issue consent gate: if no tracking issue exists, tell the user and ask whether to file one (per the target repo's workflow) or produce a self-contained handoff. Do not file silently. If an existing issue is stale, flag the difference as unpersisted delta instead of silently editing it.
  4. Compress by reference: for each piece of information — already in a durable artifact → give path/URL + status + anchor only; not persisted but required to continue → state it briefly, marked as unpersisted decision / blocker; irrelevant to the next step → omit. If artifacts conflict with observed state, name the conflict and say which evidence wins.
  5. In-progress tasks: include branch/worktree, base, dirty files, done/not-done, latest verification, failures or blockers, and the exact next step. Quote only actionable failure conclusions and commands, never whole terminal logs. Never claim unverified work as done.
  6. Safety pass: strip API keys, tokens, passwords, connection strings, credential-bearing URLs, and PII the task does not need. Keep variable names, replace values with <redacted>. If a secret looks already-leaked, say it needs rotation without repeating the value.
  7. Output a copyable handoff prompt with the applicable sections (omit empty ones): Goal / Sources of truth / Current state / Unpersisted decisions & blockers / Next action / Suggested skills. Tell the next agent to read the sources of truth before acting. Suggest only the target repo's skills that the task actually needs; do not invent gates.

Boundaries

  • Independent explorer/parallel subtasks whose spawn prompt already states goal, boundaries, and output do not need a handoff first.
  • Handoff replaces neither issue filing, context writing, implementation, verification, nor PR workflows; it only compresses state and routes the next step.

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 741 tokens

Counted across 742 of the 995 authors here whose files we hold, read 2026-08-07

  • Reference existing artifacts by path or URLin 53 of 742, across 25 files
  • Run the full test suite after integrating changesin 51 of 742, across 19 files
  • Dispatch one agent per independent problem domainin 50 of 742, across 17 files
  • Verify fixes do not conflictin 45 of 742, across 13 files
  • Include a suggested skills section in the documentin 45 of 742, across 17 files
  • Redact sensitive informationin 41 of 742, across 11 files
  • Save to the temporary directory of the operating systemin 39 of 742, across 10 files
  • Tailor the document to user-provided focus argumentsin 39 of 742, across 9 files
  • Spot check agent changes for systematic errorsin 34 of 742, across 7 files
  • Write a handoff document summarising the current conversationin 31 of 742, across 6 files
  • Assign each agent a specific scopein 23 of 742, across 8 files
  • Provide specific scope and clear goalin 23 of 742, across 5 files

Said here and by no other author read

  • mirror the user's language
  • reuse durable artifacts by reference
  • never copy existing artifacts into a new source of truth
  • return the handoff in the conversation by default
  • ensure no facts are duplicated
  • state unpersisted information briefly

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.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.