agentleFS
tethertotetherto/qvac/.agents/skills/qv-qip-review/SKILL.md

qv-qip-review

Skill697 starsChanged 18 days agoApache-2.0View on GitHub

Reviews QIP drafts for approval readiness, architectural fit, decision clarity, material trade-offs, and appropriate stakeholder detail. Use when reviewing a QIP or invoking /qv-qip-review.

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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:

    Read ../qv-qip-create/references/qip-template.md.

    line 12
  2. Very specific:

    When the proposal changes runtime, package, plugin, registry, storage, transport, security, public API, release, or deployment boundaries, …

    line 14
  3. Distinctive:

    Do not infer approval, commitments, current behavior, or implementation feasibility without evidence.

    line 16
  4. Distinctive:

    Use 600-900 words as the target and 1,200 words as a soft ceiling for the core proposal, excluding the approvers table and Appendix.

    line 34
  5. Distinctive:

    Check that the Appendix is relevant and organized, research has been synthesized rather than reproduced, external artifacts explain their …

    line 38
  6. Distinctive:

    If consultation context is provided, check coverage of the owning team lead, Lead / Architect, and any relevant cross-cutting expertise.

    line 42

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

QIP Proposal Review

Review a QIP without substituting for its human approvers. Use qv-qip-create for first drafts and qv-qip-triage to decide whether a QIP is needed.

Ground the review

  1. Read ../qv-qip-create/references/qip-template.md.
  2. Read docs/architecture/PRINCIPLES.md.
  3. When the proposal changes runtime, package, plugin, registry, storage, transport, security, public API, release, or deployment boundaries, check its system fit against docs/architecture/ARCHITECTURE.md and the relevant current repository sources.

Do not infer approval, commitments, current behavior, or implementation feasibility without evidence. Treat principle conflicts as review findings, not automatic rejection, unless the proposal hides or misrepresents the conflict.

Review criteria

Decision readiness

  • Problem explains what matters and why a decision is needed now.
  • Solution recommends one direction and states the exact approval ask.
  • Architectural responsibilities, boundaries, interactions, and rationale are concrete enough to evaluate.
  • Obvious alternatives are addressed briefly, with supporting research summarized in the Appendix when needed.
  • Consequences state positive impact and the trade-offs reviewers must accept.
  • Decision-relevant trust-boundary, compatibility, migration, and release effects are explicit.
  • Likely scope misunderstandings are excluded explicitly, and the approvers table is preserved.

An unclear or absent recommended direction or approval ask is a blocker. For another gap, first state how its answer could materially change the recommendation or why approval without it would be irresponsible. Explicitly deferred implementation detail is acceptable when the architectural direction, boundaries, responsibilities, and accepted trade-offs remain clear; require production-level detail only when the decision or its feasibility and risk claims depend on it. Do not use length or a preference for greater specificity as a proxy for readiness.

Decision-brief quality

  • Use 600-900 words as the target and 1,200 words as a soft ceiling for the core proposal, excluding the approvers table and Appendix.
  • Length above the ceiling is a suggested edit, not a blocker by itself.
  • Flag specific passages when repetition or implementation detail buries the problem, recommended direction, architectural boundary, or accepted trade-offs.
  • Recommend removing file lists and moving useful APIs, protocols, execution steps, test plans, rollout detail, failure-mode analysis, and large comparisons to the Appendix unless they directly affect approval.
  • Check that the Appendix is relevant and organized, research has been synthesized rather than reproduced, external artifacts explain their relevance, principle references explain a concrete fit or conflict, and images are linked rather than embedded as base64 data.

Consultation

If consultation context is provided, check coverage of the owning team lead, Lead / Architect, and any relevant cross-cutting expertise. The consultation note belongs outside the Canvas-ready QIP, so its absence from the QIP is not a finding. Advice is direction plus reasoning, not a vote.

Findings and severity

Separate:

  • Blockers: the decision cannot responsibly be approved. State the approval risk before using this category; examples include no clear approval ask, an unsupported material claim, an unresolved architectural contradiction, or missing impact that could materially change the decision.
  • Clarifying questions: answers would improve confidence but are not shown to prevent responsible approval or require restructuring the proposal.
  • Suggested edits: concision, organization, Appendix moves, and other improvements that do not block the decision.

Lead with findings ordered by approval risk and use line-specific references when possible:

## Blockers
- ...

## Clarifying questions
- ...

## Suggested edits
- ...

## Approval readiness
Ready | Ready with minor edits | Not ready

## Slack comment
<optional concise paste-ready comment if requested>

Say explicitly when there are no blockers. Do not rewrite the whole QIP unless asked. Keep an optional Slack comment under one screen.

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.