rlaope/oh-my-hermes / .claude/agent-skills/omh-visual-qa
omh-visual-qa
Verified[omh] Rendered UI needing a visual verdict: prepare observed-only…
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.9 KB |
| references/details.md | markdown | 2.9 KB |
| references/motion-continuity.md | markdown | 3.1 KB |
| references/playable-check.md | markdown | 2.1 KB |
| references/visual-verdict-contract.md | markdown | 4.9 KB |
references/playable-check.md
View on GitHub# Playable Check
A game or interactive scene can have a clean layout, a fine still frame, and
still not be playable: it boots to a blank canvas, ignores the key it
advertises, or freezes after the first frame. A screenshot of one frame
cannot tell these apart, so a game gets a probe of four captures.
## The four checks
1. **Boots.** The first frame, captured exactly as booted, is not blank and
not a single flat colour.
2. **Answers input.** Capture a frame before and a frame after the declared
input (a key, click, or touch). The two differ in the region the input
should move; a difference elsewhere (a clock, a particle layer) does not
count.
3. **Keeps running.** When the game has a loop, frames taken a few seconds
apart while idle differ. Background-only animation (drifting clouds, a
pulsing title) does not count; the thing that should move, moves.
4. **Clean console.** The console shows zero errors across boot and the
input. A warning is listed; an error fails the check.
## Verdicts
- All four checks backed by a capture: PASS for this lane.
- A check whose capture is missing is BLOCK, not PASS. Reading the code, or
a description of what the code should do, never passes a check.
- A check whose captures show the failure is REVISE with the smallest fix.
- A run with no WebGL context (a headless browser without GPU support) is
recorded as `environment: no_webgl`. It is not judged as a blank frame;
the probe is re-run where a context exists, and until then the checks it
blocks stay unobserved.
## Record the probe
| check | capture path | verdict |
| --- | --- | --- |
| boots | `captures/boot.png` | PASS / REVISE / BLOCK |
| answers input | `captures/before-input.png`, `captures/after-input.png` | PASS / REVISE / BLOCK |
| keeps running | `captures/idle-0s.png`, `captures/idle-5s.png` | PASS / REVISE / BLOCK |
| clean console | `captures/console.txt` | PASS / REVISE / BLOCK |
## Boundary
A probe plan is not a probe. The table is observed evidence only for rows
whose capture paths exist from the target revision; the rest stay
`prepared_not_observed`.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |