agentleFS
Sign inStart free
HarambeeAIHarambeeAI/bayes-ship/skills/ship-spec/SKILL.md

ship-spec

Skill1 starsChanged 7 months agoMITView on GitHub

Generate a structured PRD from discovery and research outputs. Includes spec review subagent with 3-iteration cap. Requires user approval before proceeding. Must run after ship-research.

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

6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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:

    Read .claude/state/CONTEXT.md and .claude/state/RESEARCH.md.

    line 10
  2. Distinctive:

    Before presenting to the user, dispatch a spec-document-reviewer subagent:

    line 27
  3. Distinctive:

    Only flags issues that would cause real problems during implementation

    line 30
  4. Distinctive:

    If issues found: revise the spec and re-dispatch the reviewer

    line 31
  5. Distinctive:

    Maximum 3 iterations: If reviewer still finds issues after 3 tries, surface the unresolved conflict to the user for a decision.

    line 32
  6. Very specific:

    Update progress.json: set phases.spec.status to "approved".

    line 39

≈ “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

Spec — PRD generation with automated review

Pre-condition

Read .claude/state/CONTEXT.md and .claude/state/RESEARCH.md. Both must exist.

Process

  1. Read CONTEXT.md + RESEARCH.md.
  2. Generate a structured PRD with these sections:
    • Introduction and problem statement
    • Goals (measurable)
    • User stories (US-001 format). Each story must be completable in 2-5 minutes of AI work. If a story cannot be described in 2-3 sentences, it is too big — split it.
    • Acceptance criteria per story (verifiable, not vague). Every story includes "Typecheck passes" and "All tests pass" as final criteria. UI stories add "Verify changes work in browser."
    • Non-goals (what this version will NOT do)
    • Technical considerations (informed by research)
    • Data model sketch
  3. Order stories by dependency: schema → backend → frontend → integration → UI polish.

Spec review subagent

Before presenting to the user, dispatch a spec-document-reviewer subagent:

  • Provide ONLY the spec file path and CONTEXT.md (never full session history)
  • Reviewer checks for: completeness, internal consistency, clarity, scope creep, YAGNI violations
  • Only flags issues that would cause real problems during implementation
  • If issues found: revise the spec and re-dispatch the reviewer
  • Maximum 3 iterations. If reviewer still finds issues after 3 tries, surface the unresolved conflict to the user for a decision.
  1. Present the reviewed PRD to the user in digestible chunks:
    • Goals section first (approve/adjust)
    • User stories summary (approve/see details/adjust)
    • Non-goals (approve/adjust)
  2. Save approved PRD to .claude/state/PRD.md.
  3. Update progress.json: set phases.spec.status to "approved".
  4. Output: "Ready. Run /bayes-ship:ship-architect to define the system architecture."

Gate

The user MUST explicitly approve the PRD before /bayes-ship:ship-architect can run. No implicit approval.

Story sizing reference

Right-sized: Add a database column and migration. Add a single UI component. Update a server action. Add a filter dropdown.

Too big (must split): "Build the dashboard" → schema, queries, UI components, filters. "Add authentication" → schema, middleware, login UI, session handling.

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