agentleFS
Sign inStart free
dagarre00dagarre00/claude-code-template/.claude/skills/dispatch-handoff/SKILL.md

dispatch-handoff

Skill1 starsChanged 27 days agoMITView on GitHub

Conductor-only. How to hand one role's dispatch to the human as a file they run in another harness (a fresh Claude Code session, Antigravity/agy, or another CLI) and verify the returned result through the same checks as a native dispatch.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    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
  1. Very specific:

    Take back: Read .handoff/<slug>-<role>-report.md (or the pasted message).

    line 25
  2. Very specific:

    Clean up: Delete both .handoff/<slug>-<role>-*.md files once accepted or discarded so later reviewers cannot read them.

    line 28

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Dispatch Handoff

Hands a role's dispatch to the human as a file to run in an external harness when dispatch mode is handoff (subagent-dispatch § Dispatch mode). The brief, checkout, baseline check, and Red/Green proof are identical to a native dispatch.

Read first

  • .claude/agents/<role>.md — extract model, effort, tools, and skills:.
  • handoff-template.md — template for .handoff/<slug>-<role>-handoff.md.

Steps

  1. Baseline: Require a clean tree and record git rev-parse HEAD and git status --porcelain. No other writer touches the checkout until the handoff returns.
  2. Paths: <slug> is the entity slug or command name (review, wiki). Write .handoff/<slug>-<role>-handoff.md and expect .handoff/<slug>-<role>-report.md (delete any pre-existing report file first).
  3. Fill handoff-template.md:
    • Brief: the exact four-line Task / Inputs / Scope / Return block (rule 10: no conductor reasoning, files passed by path).
    • Reading list: .claude/rules/behavioral.md, .claude/rules/workflow.md, subagent-contract, .claude/agents/<role>.md, and each skill in the role's skills: list in order.
    • Limits & model: the role's tools, access mode (read-only vs write), model, and effort (note: Antigravity --effort caps at high).
  4. Hand over: Tell the human the handoff file path and launch commands, and wait for "done" (with the harness/model used) or a pasted report. Make no edits while waiting.
  5. Take back: Read .handoff/<slug>-<role>-report.md (or the pasted message). If missing or partial, ask the human to continue the external session once (partial again → human-checkpoint). Then run subagent-dispatch § Wait, then check what it touched step 2 (and § Prove and commit a developer case for developer).
  6. Send back: Give the human the deciding output tail to paste into the same external session (or write a fresh handoff file with "these files are the previous attempt" if starting a new session).
  7. Record harness: For reviewer roles, record the external harness and model per finding-disposition § Record the yield.
  8. Clean up: Delete both .handoff/<slug>-<role>-*.md files once accepted or discarded so later reviewers cannot read them.

Anti-patterns

  • Leaking conductor reasoning into the handoff file.
  • Skipping the baseline diff check (external harnesses do not enforce .claude/agents/ tool restrictions).
  • Running a reviewer in a session that already saw the author's context.
  • Leaving stale .handoff/ report files on disk.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free