agentleFS
June74June74/agentic-workflow/skills/work-order/SKILL.md

work-order

Skill0 starsChanged 3 months agoMITView on GitHub

Use for every downward dispatch from the Scrum Master to a team. Defines the contract a team receives before starting work; a dispatch without one is a vague prose handoff.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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:

    Every sys_session_send that assigns work carries a work order.

    line 12
  2. Distinctive:

    If the criteria were derived rather than given by the PM, say so and point at the Decision Record where they were written down.

    line 24
  3. Very specific:

    Put the work order's ID in the title (WO-0007-conflict-api) so uniqueness is structural rather than a naming habit.

    line 57

≈ “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

Work order

Vague prose handoffs are the most common way multi-agent systems produce confident, useless work. A team that does not know what "done" means will pick a definition, and it will be the one that is easiest to reach.

Every sys_session_send that assigns work carries a work order. Use templates/work-order.md.

Required fields

Goal — one sentence. What changes in the product, not what files get edited.

Acceptance criteria — a checklist, each item independently verifiable, each naming how it will be verified. "Conflicts are detected" is not a criterion. "Given two events overlapping by >=1 minute in the same domain, the conflict list contains both, verified by pytest tests/test_conflicts.py -q" is.

If the criteria were derived rather than given by the PM, say so and point at the Decision Record where they were written down.

Constraints — what must NOT change. Files out of scope, interfaces that are frozen, performance or compatibility budgets, style conventions to follow.

Interfaces to respect — the contracts this work sits behind: the UX interface contract, an API shape, a schema. Name the file, not the concept.

Definition of done — what returns. A PR number, a tag, a findings file, a document. Name the artifact.

Titles

Set title on the dispatch to the WORK, never the vendor or harness.

  • Good: conflict-detection-api, review-briefing-privacy, explore-sync-flake
  • Bad: codex, claude_code, worker, agent-2

The title is what the PM sees in the session list. A list of vendor names tells them nothing about what is running.

A title is also a continuation key. sys_session_send does spawn-or-continue on it: reuse a title and you continue that conversation rather than starting a new one. A title must therefore be unique to one task and one recipient, and reused only to deliberately continue that exact conversation.

Two dispatches sharing a title — or a title recycled from earlier in the run — can bind a session to the wrong conversation entirely. That has happened here: a dispatched design brief came back carrying a byte-identical copy of an earlier liveness probe, and the agent answered the probe faithfully while the real question was never asked. See docs/incidents/11.

Put the work order's ID in the title (WO-0007-conflict-api) so uniqueness is structural rather than a naming habit. And validate the return: a work order that comes back without addressing its own acceptance criteria was misbound, not failed — re-dispatch on a fresh title rather than treating it as a bad delivery.

Scope discipline

One work order, one scoped task. If a work order needs three unrelated things, it is three work orders. Teams that receive bundled scope return bundled evidence, and bundled evidence cannot be checked.

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.