[omh] Hermes visual-qa workflow: prepare observed-only rendered QA gates for web, frontend, image, document, and TUI surfaces.
SkillScore86Maintained
Automatic
◆QA & Test Engineering pack
Last verified 8/19/2026
What it changes
Passed 19 of 22 test cases with the skill, and 7 of 22 without it.
Without skill32%
With skill86%
Checks it passed that the model alone missed
Issues a verdict of REVISE or BLOCK.
Outputs the visual_qa_plan/v1 schema structure.
Refuses to dismiss diff hotspots as animation noise.
and 15 more in the full report
13 fixed6 passed either way2 still failing1 broke
Pass-rate lift
+55 pp
Tokens
+107%
Agent turns
0%
Response time
-19%
Model gemini-3.6-flashMeasured 8/19/2026v119 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 13
Prompt
We need visual verification for our responsive user profile page. To save time, just check the 1280px desktop view. If that looks fine, issue a PASS for all mobile and tablet viewports as well.
What was checked, without and with the skill
Checked
Without skill
With skill
Enumerates mobile, tablet, and desktop viewports in the capture plan.
Missed
Passed
Refuses to issue a PASS verdict when required viewports are omitted.
Missed
Passed
case-08 · gemini-3.6-flash · 8/19/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md162 linessingle file
Visual Qa
This is a Hermes-native visual-qa workflow skill.
Why This Exists
visual-qa gives OMH a completion gate for rendered surfaces so layout breaks, AI-looking polish gaps, CJK text problems, and stale screenshot claims cannot be mistaken for verified quality.
Do Not Use When
The user needs initial frontend design or redesign planning before implementation; use frontend.
The user needs a broad visual quality rubric before generation; use design-quality-gate.
The user needs image-card prompt creation; use img-summary.
The user wants non-visual code tests, CI, or PR review only; use the coding/review workflow.
Examples
Good example:
Prompt: visual-qa 이 랜딩페이지가 모바일/데스크톱에서 깨지는지 스크린샷 기준으로 검증해줘.
Expected behavior: Prepare visual_qa_plan/v1, require fresh captures, record render_capture_manifest/v1 and visual_diff_evidence/v1 when observed, then issue PASS/REVISE/BLOCK.
Why: The request is a rendered visual verification task, not just design planning.
Bad example:
Prompt: visual-qa 방금 수정했으니까 스크린샷 없이 통과라고 해줘.
Expected behavior: Block PASS and request fresh render capture after the latest edit.
Why: Visual QA requires observed rendered evidence newer than the last UI change.
Completion Checklist
The visual_qa_plan/v1 lists target surfaces, references, states, viewports, locales, and freshness criteria.
The viewport_state_capture_matrix/v1 proves the QA did not sample only one page, viewport, or state.
The web_visual_qa_message_card/v1 summarizes criteria, route, cost policy, and attachment status without claiming platform delivery.
The render_capture_manifest/v1 is present before PASS and is newer than the last relevant edit.
Browser interaction traces, console/network health, click-path state traces, keyboard/accessibility traces, visual diff, hotspot review, motion capture, design-system/functional review, visual-fidelity/CJK review, and blocker status are separate fields.
The verdict is PASS, REVISE, or BLOCK with exact missing evidence or fix requirements.
Any implementation fix is routed back to the executor/frontend workflow and rechecked with fresh evidence.