agentleFS
Sign inStart free
mepottsmepotts/agent-operating-kit/skills/sprint-spec/SKILL.md

sprint-spec

Skill0 starsChanged 43 days agoMITView on GitHub

Write a sprint spec before any code (outcome, file boundaries, risk tier, acceptance checks, isolation). Use when the user wants to plan, scope or spec work for an AI coding agent, asks for a definition of done or acceptance criteria, or is about to start a feature, fix, refactor or migration.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    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:

    Pending criteria: If runtime proof is not possible yet, list it as pending with a reason.

    line 17
  2. Distinctive:

    No code before the outcome and the tier are agreed.

    line 24
  3. Distinctive:

    Keep it under two pages. Link depth instead of pasting it.

    line 26
  4. Very specific:

    No repository in scope, or an empty one: do not invent paths or commands.

    line 27

≈ “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

Sprint spec

Specify before code. A spec is finished when someone who did not write it could tell, from files alone, whether the work is done.

Template: ${CLAUDE_PLUGIN_ROOT}/templates/SPRINT.md. Command: /agent-operating-kit:sprint. Tiers: the risk-tiers skill.

Checklist

  1. Outcome. One to three sentences of observable behavior, plus what is out of scope.
  2. Risk tier (0-4). The highest tier touched wins. It sets review depth and who signs off.
  3. File boundaries. Owned paths, shared files that need coordination (schemas, shared types, CI, AGENTS.md), forbidden paths. One writer per file across parallel work.
  4. Acceptance. For each outcome, an executable check named exactly (a command or a test title) and what it must show. Cover error, cancel and persistence states where they apply. Add the end-to-end check on the real run target, not only a typecheck.
  5. Pending criteria. If runtime proof is not possible yet, list it as pending with a reason. A mock never closes a runtime criterion.
  6. Isolation. Own worktree and branch, the data target set explicitly (production credentials never in a workspace), ports and shared resources reserved.
  7. Roles and effort. Who implements and who reviews (never the same agent). Model and effort per workstream, with a one-line reason.
  8. Stop conditions. Unrecognized files or branches, partial failures, unfamiliar response shapes, errors you did not cause: stop and report.
  9. Approvals. List every irreversible action the work might need. Each needs approval of the exact change.

Rules

  • No code before the outcome and the tier are agreed.
  • "Done" is written as checks, not adjectives. If a criterion cannot fail, rewrite it.
  • Keep it under two pages. Link depth instead of pasting it.
  • No repository in scope, or an empty one: do not invent paths or commands. Name file boundaries by role (for example "route handlers", "shared middleware"), mark each confirm against the repo, and say what you need from the human.
  • Save it at the spec path named in the project's AGENTS.md (default sprints/<slug>.md), then stop and ask the human to confirm outcome and tier.

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