Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/valkor-ai/loom/visual-qanpx skills add valkor-ai/loom --skill visual-qagit clone --depth 1 https://github.com/valkor-ai/loomWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/valkor-ai/loom/visual-qa)<a href="https://agentmods.dev/skills/valkor-ai/loom/visual-qa"><img src="https://agentmods.dev/badge/skills/valkor-ai/loom/visual-qa.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00045 | $0.00371 |
| Opus 5 | $0.00023 | $0.00186 |
| Sonnet 5 | $0.00009 | $0.00074 |
| Haiku 4.5 | $0.00005 | $0.00037 |
Grade A, and why
visual-qa scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Godot Visual QA
Visual QA is evidence review, not a style contest. Compare the capture with the task's acceptance criteria and report only issues that affect the requested behavior, visibility, composition, or operation.
Capture first
Use the screenshot skill or the Godot MCP screenshot tool. Wait for the scene
to render before capturing. For motion, capture a reference frame followed by
several frames at a fixed interval so movement can be checked.
Review checklist
- The intended scene is running and the viewport is not black or empty.
- The camera shows the requested objects at a usable scale.
- Meshes, textures, lighting, and materials are present rather than placeholders when the task requires them.
- Objects that should move or interact visibly change state between frames.
- UI text and controls fit their containers and do not overlap the scene.
- Collision shapes, bodies, and navigation behave consistently with the visual geometry when debug overlays are enabled.
- Errors visible in the capture or reported by the runtime are recorded as evidence, not guessed away.
Result format
Record a short result in the Loom task evidence or handoff:
Verdict: pass | warning | fail
Evidence: <capture paths>
Observed: <what the frames show>
Issues: <file and location, or none>
Use fail for an acceptance-blocking issue, warning for a non-blocking
defect, and pass only when the supplied criteria are visible in the capture.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 47 lines · 45 tokens per session scan A fc3102c60e97
visual-qa is a skill published in the GitHub repository valkor-ai/loom (949 stars, last pushed 18d ago), licensed Apache-2.0. It adds 45 tokens to every session and 371 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
threejs-qa-release
Verify and release Three.js browser games. Combines playtest QA, automated bot playtests, mobile/responsive checks, production builds, preview verification, static-hosting base paths, debug gating, bundle review, screenshots, visual test harness decisions, packaged canvas-pixel inspection with measured metrics…
hearth-playtest
Let the engine hunt bugs for you — bot playtesting via hearth sweep. Seeded bot policies (mash/idle/wander/seek) play a scene headlessly across many seeds and report softlocks, crashes, stuck states, and unmet objectives as a compact evidence report; objectives double as executable acceptance criteria; a failing seed…
unity-testing-verification
Use when running Unity tests, authoring or linting playtest DSL, verifying Play Mode behavior, or reporting gameplay and visual claims from evidence.
smashup-faction-implementation
Smash Up 派系玩法实现、旧派系参考、测试、E2E 与审计 workflow。.
godot-input-testing
Simulate keyboard, mouse, and input-action events in a running Godot game through the godot-mcp CLI for automated UI and gameplay testing. Use when testing controls, verifying UI navigation, automating gameplay sequences, simulating key presses or mouse clicks, or checking that input actions respond correctly.
grave-poi-death-test
通过真实游戏模拟验证角色死亡、墓碑 POI 创建、地图渲染和点击详情。用于需要测试寿元死亡或其他游戏内死亡流程、墓碑 POI，且不得直接调用 handledeath 的任务。.