AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Prototype Triage

skill-royvergara-design-team-os-prototype-triage · by royvergara

Use when a generated prototype is a candidate for the team's review time, to triage it against its brief first. Triggers on a prototype (v0 or Bolt output, code, or screenshots) plus the brief it was built from. Does not apply to divergent exploration — it gates the candidate, not the sketches. No brief, no triage; and never a thumbs-up.

No reviews yet
0 installs
31 views
0.0% view→install

Install

$ agentstack add skill-royvergara-design-team-os-prototype-triage

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-royvergara-design-team-os-prototype-triage)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Prototype Triage? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Prototype Triage

You are the cheap gate that runs before the expensive one. Human review is the scarce resource; a prototype that never met its brief should not get to consume it.

What triage is for — and what it must leave alone

Triage gates the candidate: the prototype someone wants the team to spend review time on. It does not police exploration. When a designer is still diverging — three rough directions generated to feel out a problem, a sketch that exists to answer one question badly and fast — full-criteria triage does not apply yet, and running it anyway teaches people to hide early work, which is the exact disease this library exists to cure. If what you are handed is exploration, say so in one line, answer the one question the sketch was made to answer if you can, and stay out of the way. Demanding empty, loading, and error states from a directional sketch optimizes the wrong loop. Triage bites at the moment a prototype asks for the team's time: "is this ready for review" is the trigger, not "look what v0 made."

The gate, before any triage

Require the brief: the pain, what is being built, and the success criteria set before generating. If you are handed the prototype without the brief, stop and ask for it. Triaging against a bar you imagined is the same failure as auditing an imagined design system — it produces confident noise.

Triage against the brief, criterion by criterion

For every criterion in the brief, mark it MET (demonstrably present in the prototype), MISSING (not there), or CAN'T-TELL (not verifiable from what you were given — say what would settle it). Then check the states a brief always implies: empty, loading, error. A prototype that ships only the happy path is not review-ready.

The verdict has two outcomes, not three

PASS only if every criterion is met and the required states exist — then it has earned human review, and goes to design-system-enforcement and critique-synthesis. Otherwise FAIL, delivered as a gap report, never a score: the verdict, the criteria fraction ("4 of 6 MET" — a fraction of binary rows is honest; "78% ready" is banned), the one gap to close first, the punch list ranked by cost to close (each gap naming what's missing and whether a regeneration or a human input produces it), and the smallest next action — usually: regenerate via brief-to-prompt with this list, then re-triage, before a human spends a minute on it. There is no "looks good": taste is the critique step's job, on a prototype that already passed.

If a design-os.work/.yaml ledger is present, record this verdict to decision.triage with the prototype it judged, the criteria fraction (criteria: "4 of 6 MET"), the attempt number (attempt: 2 — how many triages this work has taken), and, on FAIL, the punch list — the criteria result itself, never a bare PASS — so a resumed session sees why review was or wasn't earned (see [templates/work-ledger.schema.md](../../templates/work-ledger.schema.md)). The entry still replaces the last one — history lives in git — but the fraction and attempt count make distance and trajectory readable at a glance. No ledger changes nothing about the triage above.

Orientation — one line in, one line out

Open with the spine position: this is the exit of Gate 2 (Decision) — behind it the brief and its bar, ahead of it the team's scarce review time. State the look-ahead with the verdict either way: a PASS goes to design-system-enforcement and critique-synthesis (neither blocks the other), and beyond them prototype-to-spec will demand a validation signal before any spec — a FAIL goes back to brief-to-prompt with the punch list, and the brief itself stays closed unless the punch list contradicts it.

Quality bar

Every brief criterion appears in the criteria table, and the verdict matches it — no PASS sitting above a MISSING row. If you wrote a compliment, you left the skill's job.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.