agentleFS
58254727-blip58254727-blip/codex-task-control-tower/skills/task-handoff/SKILL.md

task-handoff

Skill1 starsChanged 3 months agoMITView on GitHub

Use when a Codex task must continue in another task or session with verified context, explicit boundaries, and a compact privacy-safe handoff.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 7

    Distinctive

    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
  1. Distinctive:

    Create a continuation record from verified work, not a transcript dump.

    line 8
  2. Distinctive:

    Verified completed: only outcomes supported by tests, diffs, commits, logs, or receipts.

    line 16
  3. Distinctive:

    Do not include raw conversation by default.

    line 24
  4. Very specific:

    Never invent a completed item, command result, path, commit, timestamp, or blocker.

    line 26
  5. Distinctive:

    Read the latest verified artifact, status, and boundary notes.

    line 30
  6. Distinctive:

    Remove everything not needed for the destination task's next safe action.

    line 32
  7. Distinctive:

    Run a final privacy check before sharing or saving the handoff.

    line 34
  8. Distinctive:

    Do not silently broaden scope, authorize external actions, or transfer secrets.

    line 36

“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.

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

Task Handoff

Create a continuation record from verified work, not a transcript dump.

Required Fields

Include these sections in this order:

  1. Objective: the concrete outcome still being pursued.
  2. Current scope: files, modules, or behavior currently in bounds.
  3. Verified completed: only outcomes supported by tests, diffs, commits, logs, or receipts.
  4. Latest evidence: the newest inspectable evidence and when it was produced.
  5. Blocker or user action: the exact dependency, approval, or missing input.
  6. Frozen boundaries: behavior, data, files, or environments that must not change.
  7. Next safe step: one action the destination task can perform immediately.

Sanitization

Do not include raw conversation by default. Remove credentials, authentication material, private data, internal addresses, personal contact details, and unrelated history. Include identifiers only when explicitly requested and necessary for continuation. Prefer generic labels in examples.

Never invent a completed item, command result, path, commit, timestamp, or blocker. Mark unknown facts as unverified and state what evidence would resolve them.

Procedure

  1. Read the latest verified artifact, status, and boundary notes.
  2. Separate facts from assumptions.
  3. Remove everything not needed for the destination task's next safe action.
  4. Populate the required fields using short bullets.
  5. Run a final privacy check before sharing or saving the handoff.

Do not silently broaden scope, authorize external actions, or transfer secrets. A handoff communicates state; it does not grant permission.

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.