agentleFS
rpraharajrpraharaj/ai-engineer/skills/write-dev-tasks/SKILL.md

write-dev-tasks

Skill1 starsChanged 15 days agoMITView on GitHub

Break a PRD or technical design into ordered, implementable tasks with scope and yes/no acceptance. Use when the user asks for tickets, a task breakdown, detailed dev tasks, or a work plan. Do not write product code. Do not invent a backlog for a one-file change.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 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.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not implement. Do not open tracker issues unless the user asked and a tracker CLI this repo already uses works.

    line 14
  2. Very specific:

    If there is only one task and it is already small, stop and run shape-work.

    line 25
  3. Very specific:

    Read AGENTS.md if present.

    line 27
  4. Very specific:

    Use the first that exists: docs/tasks/, docs/dev/, a path AGENTS.md names, or a sibling of the design/PRD.

    line 31
  5. Very specific:

    If none: write in the reply, and offer to create docs/tasks/ in this repo's markdown style.

    line 33

≈ “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

Write dev tasks

Turn an agreed what/how into an ordered list of slices another session can pick up. This is not a PRD, not a design, and not the implementable contract for a single slice (shape-work).

Do not implement. Do not open tracker issues unless the user asked and a tracker CLI this repo already uses works.

Is this the right skill?

SignalDo this instead
Product why still openwrite-prd
How we will build it still openwrite-design
Exactly one small sliceshape-work
Start codingimplement-change

If there is only one task and it is already small, stop and run shape-work. Do not produce a one-item backlog.

Read AGENTS.md if present. Read the PRD / design / architecture the user pointed at. If none exists and the work is large, say so and offer write-prd or write-design first.

Find the home

Use the first that exists: docs/tasks/, docs/dev/, a path AGENTS.md names, or a sibling of the design/PRD.

If none: write in the reply, and offer to create docs/tasks/ in this repo's markdown style.

Name the file like existing task lists. If there are none, docs/tasks/<slug>.md.

Write

Copy headings from references/dev-tasks-template.md in this skill directory. Keep every heading; write n/a when a section does not apply. Do not invent extra sections. If that file is missing, use the same headings: Goal, Prerequisites, Order, Task N (Kind, Depends on, Traces to, In/out of scope, Acceptance, Likely files, Prove, Notes, Next skill), Out of this batch, Batch done, Risks, Next.

Reuse agreed upstream decisions and acceptance checks by reference; add only the detail needed for this slice. Do not repeat a design or re-open settled choices. Investigation tasks may accept a bounded evidence report rather than a behavior change.

Each task:

  • One kind. One next skill.
  • Acceptance is yes/no a stranger can verify. For a behavior change, at least one check would fail today. For an investigation, specify the question, time/effort bound, and evidence artifact instead.
  • Likely files are real paths, or you say you could not find them.
  • Traces to names the PRD requirement or design slice when those docs exist.
  • Small enough that shape-work can tighten it or implement-change can take it if it is already clear.
  • No "and also refactor the world" inside a feature task. That goes to Out of this batch or its own task.

Order is dependency-real (schema expand before readers, flag before default-on). Parallel only when there is no shared contract.

Quality

  • Task 1 is the smallest slice that proves the approach.
  • Later tasks do not assume code Task 1 has not shipped.
  • Batch done would fail today.
  • Count is as small as the risks require. Do not split by file to look thorough.

Done

  1. Path or the full text in the reply
  2. Task count and the first task title
  3. Next skill for Task 1

Do not implement the list. Follow Task 1's recorded kind and next skill: debug for defects, data-change for migrations, write-docs for docs, or verify-ui for UI verification. For a change use shape-work, or implement-change when already clear. For a spike, run a bounded investigation and report evidence without implementing a speculative solution.

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.