agentleFS
markoblogomarkoblogo/abvx-agent-skills/skills/idea-to-ship-gates/SKILL.md

idea-to-ship-gates

Skill16 starsChanged 5 months agoMITView on GitHub

Route an idea, vague feature request, or agent-built product change through delivery gates from intent to release. Use when work needs more discipline than a readiness check: clarify the spec, slice the work, protect architecture boundaries, gather proof, converge through review, and prepare release without turning the task into a heavyweight process.

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 10 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Start at the first missing gate:

    line 28
  2. Distinctive:

    Architecture gate: check boundaries, data flow, ownership, and blast radius before broad edits.

    line 33
  3. Very specific:

    Use rapid-grilling or doc-grounded-grilling for the intent gate.

    line 41
  4. Very specific:

    Use spec-to-prd when the spec is not yet durable enough.

    line 42
  5. Very specific:

    Use plan-to-issues for slice planning.

    line 43
  6. Very specific:

    Use architecture-deepening-review or system-zoom-out for architecture risk.

    line 44
  7. Very specific:

    Use delivery-preflight-gate before long or risky execution.

    line 45
  8. Very specific:

    Use test-driven-execution, minimal-diff-builder, or phase-spec-execution during implementation.

    line 46
  9. Very specific:

    Use browser-verification, web-quality-audit, or repo-specific checks for proof.

    line 47
  10. Very specific:

    Use overengineering-review, delivery-baseline-audit, and recovery-loop-3strike for convergence.

    line 48
  11. Very specific:

    Use agent-learning-layer-triage after repeated failures or reusable delivery lessons.

    line 49
  12. Distinctive:

    Do only the work needed for that gate.

    line 56
  13. Distinctive:

    Run the smallest proof command or inspection that can falsify success.

    line 57
  14. Distinctive:

    Remove accidental scope before advancing.

    line 63
  15. Distinctive:

    Do not create process artifacts for tiny edits.

    line 67
  16. Distinctive:

    Do not skip from idea to implementation when acceptance criteria are unclear.

    line 68
  17. Distinctive:

    Do not turn release prep into proof.

    line 69
  18. Distinctive:

    Do not keep looping on taste, preference, or ambiguous product calls without user input.

    line 70
  19. Distinctive:

    Do not install external skill packs just because one pattern is useful.

    line 71

≈ “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

Idea To Ship Gates

Turn a vague idea into a shipped, reviewable slice. Use this as a router over existing skills, not as a replacement for implementation judgment.

Core Principle

Readiness loops make a repo interpretable. Delivery loops make a change shippable.

The minimum useful delivery path is:

intent -> spec -> slices -> proof -> review -> release

Advance only when the current gate has enough evidence for the next one.

Gate Router

Start at the first missing gate:

  1. Intent gate: clarify the user-visible outcome, non-goals, constraints, and risk level.
  2. Spec gate: produce a small spec or PRD only when the idea is still ambiguous.
  3. Slice gate: split work into thin end-to-end slices with acceptance criteria.
  4. Architecture gate: check boundaries, data flow, ownership, and blast radius before broad edits.
  5. Execution gate: implement the current slice with the narrowest useful verification loop.
  6. Proof gate: collect commands, screenshots, artifacts, or review notes that show the slice works.
  7. Convergence gate: review the diff, remove accidental scope, fix blockers, and repeat only on concrete findings.
  8. Release gate: prepare changelog, version, deployment, or PR handoff only after proof survives review.

Skill Routing

  • Use rapid-grilling or doc-grounded-grilling for the intent gate.
  • Use spec-to-prd when the spec is not yet durable enough.
  • Use plan-to-issues for slice planning.
  • Use architecture-deepening-review or system-zoom-out for architecture risk.
  • Use delivery-preflight-gate before long or risky execution.
  • Use test-driven-execution, minimal-diff-builder, or phase-spec-execution during implementation.
  • Use browser-verification, web-quality-audit, or repo-specific checks for proof.
  • Use overengineering-review, delivery-baseline-audit, and recovery-loop-3strike for convergence.
  • Use agent-learning-layer-triage after repeated failures or reusable delivery lessons.

Convergence Loop

Repeat this loop until the release gate passes or a blocker is honest:

  1. State the current gate and acceptance criteria.
  2. Do only the work needed for that gate.
  3. Run the smallest proof command or inspection that can falsify success.
  4. Classify the result:
    • advance: proof is sufficient for the next gate;
    • repair: one concrete blocker remains inside the current gate;
    • reslice: the slice is too large or crosses too many boundaries;
    • handoff: progress now needs user input, missing access, or a product decision.
  5. Remove accidental scope before advancing.

Guardrails

  • Do not create process artifacts for tiny edits.
  • Do not skip from idea to implementation when acceptance criteria are unclear.
  • Do not turn release prep into proof; release notes are not verification.
  • Do not keep looping on taste, preference, or ambiguous product calls without user input.
  • Do not install external skill packs just because one pattern is useful; extract the smallest reusable gate.

Final Report

Report the gates completed, evidence collected, gates intentionally skipped, and the next release or handoff action.

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.