[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79FC8C44PHWM7XCQQQAVNZA":3,"skill-same:01M79FC8C44PHWM7XCQQQAVNZA":57},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":16,"sourceHtmlUrl":17,"categories":18,"keywords":19,"publishedAt":20,"linkOnly":21,"markdownContent":22,"markdownBody":23,"frontMatter":24,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M79FC8C44PHWM7XCQQQAVNZA","screenshot-review",null,"Review image shots (screenshots, snapshot baselines, montages, renders) two ways — critique mode spawns an unprimed sub agent as a second set of eyes, MANDATORY before declaring any user-reported visual bug fixed or claiming a visual change verified; preview mode opens shots in ONE macOS Preview window when the user asks to open \u002F show \u002F pull up \u002F review shots, or right after producing visual artifacts the user should eyeball before signing off.","BumpyClock\u002Fagent-templates","BumpyClock","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1883824?v=4","01M79EW0FDP9MFXH1FKH800S7T","Aditya Sharma","01M79EW0G27TTKZWTFQ8W5HVB4",2,"MIT","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fscreenshot-review\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fscreenshot-review\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: screenshot-review\ndescription: Review image shots (screenshots, snapshot baselines, montages, renders) two ways — critique mode spawns an unprimed sub agent as a second set of eyes, MANDATORY before declaring any user-reported visual bug fixed or claiming a visual change verified; preview mode opens shots in ONE macOS Preview window when the user asks to open \u002F show \u002F pull up \u002F review shots, or right after producing visual artifacts the user should eyeball before signing off.\n---\n\n# Screenshot Review\n\nTwo modes for putting eyes on visual artifacts. Pick by audience:\n\n- **Critique** — agent eyes: spawn an unprimed sub-agent to find defects before\n  you accept visual work. Mandatory before declaring a visual bug fixed.\n- **Preview** — user eyes: open the review-worthy shots in a single macOS\n  Preview window so the user can flip through them.\n\nThey compose: run the critique first, then preview the surviving candidate\nshots for the user's sign-off. This skill is for visual defects and human\nreview, not pixel metrics; pair with `compare-screenshots` when you also need\nnumbers.\n\n## Critique mode\n\nUse an unprimed sub-agent as a second set of eyes before accepting visual work.\n\n### Workflow\n\n1. Capture or locate the exact PNGs\u002FGIF frames under review.\n2. Create tight 2x-4x crops for every key feature under judgment, plus the full\n   screenshot for context. Crop selected units, city\u002Ftown stacks, flags\u002Fpoles,\n   shadows, selection rings, labels\u002Ficons, roads, terrain features, water, and\n   any artifact-prone area. If the complaint is about \"too faint\", \"wrong\n   order\", or \"not in perspective\", the crop is mandatory.\n3. Spawn one fresh explorer with `fork_context: false`; pass only the full\n   images, the crops, and a short neutral task. Do not include the main thread\n   history, implementation details, or expected answer.\n4. Ask for concrete visible defects with confidence levels. Name likely risk\n   categories: unit\u002Fprop depth ordering, layering, shadows, selection-marker\n   contrast, ground-plane perspective, flag\u002Fpole attachment, label style and\n   icon readability, blur, scale, lighting, artifacts, missing models, terrain\n   feature readability, roads, water, and overall scan readability.\n5. Compare the sub-agent's critique against your own inspection. Treat overlap\n   as high-priority evidence. Treat novel high-confidence findings as bugs to\n   inspect, not as taste notes to dismiss.\n6. Record actionable findings in the spec, visual report, or next task plan\n   before claiming the screenshot is accepted.\n\n### Sub-Agent Prompt\n\nUse this shape, replacing the bracketed surface and attaching local images:\n\n```text\nFresh visual critique task. You have no project backstory and should only\ninspect the supplied screenshots and crops. First inspect the full screenshot\nfor context, then inspect each crop at zoomed scale. Look for concrete\nvisual\u002Flayout defects in [surface], especially unit\u002Fprop depth ordering,\nlayering, shadows, selection-marker contrast, ground-plane perspective,\nflag\u002Fpole attachment, label style\u002Ficons, blur, scale, lighting, artifacts,\nmissing models, terrain feature readability, roads, water, and scan\nreadability. Do not assume these are correct. Return a concise list of issues\nyou can see, with confidence and whether the issue is visible in the full image,\nthe crop, or both.\n```\n\nSpawn config:\n\n- `agent_type`: `explorer`\n- `fork_context`: `false`\n- attach screenshots as `local_image` items\n- omit model overrides unless the user explicitly requests one\n\n### Critique rules\n\n- **Mandatory before \"fixed\":** never declare a user-reported visual bug fixed\n  on your own inspection — your eyes are primed by the fix you just made. Run\n  the unprimed critique on the candidate shot first; \"mild residue\" you are\n  tempted to wave through is exactly what it exists to catch. (Recorded\n  failure: a \"fixed\" sky that an unprimed agent identified as the terrain\n  mesh's underside filling the entire sky region.)\n- **Reproduce the reporter's framing.** When the user supplied a screenshot,\n  the critique must include a capture at that framing (same camera\u002Fzoom\u002Fspot,\n  or as close as reproducible) — a defect that lives at their framing can be\n  invisible at yours. Your chosen probe framing is a supplement, never the\n  substitute.\n- **Prove the change is real before critiquing it.** Byte\u002Fpixel-diff the\n  candidate against the pre-change baseline first: a critique of an unchanged\n  image \"verifies\" a no-op. (Recorded failure: a palette pass that never\n  reached the production render path — before\u002Fafter were byte-identical and\n  only the diff caught it.)\n- Never tell the sub-agent the defect you expect it to find.\n- Use the current candidate screenshot, not a stale report or baseline image.\n- Do not rely on full-page report scale for small visual features. Attach\n  crops around the exact features a player would read: selected army\u002Fcity,\n  label\u002Ficon clusters, flags, shadows, ring edges, road crossings, terrain\n  feature patches, water labels, and suspicious debug\u002Fartifact regions.\n- If the sub-agent says a crop reveals an issue that is weak or invisible in\n  the full shot, treat it as a real usability defect when the player can zoom\n  to that scale in-game.\n- For animation, attach a short set of deterministic still frames first; GIFs\n  are useful for human review, but still frames make specific defects easier to\n  name.\n- A passing sub-agent critique does not replace direct inspection by the main\n  agent or screenshot regression gates.\n- If the sub-agent catches an issue the main agent missed, add that failure mode\n  to the relevant feature plan or visual checklist immediately.\n\n## Preview mode\n\nPut the review-worthy images in front of the user in a single Preview window\nthey can flip through.\n\n### Workflow\n\n1. **Curate the set** — the images that answer \"is this right?\": the\n   just-changed or just-blessed baselines, the integrated result, and the\n   reference image when there's something to compare against. Don't dump every\n   PNG in the folder; pick the smallest set that lets the user judge the\n   change. If the user named specific shots, open exactly those.\n2. **One window** — open them with a single `open -a Preview \u003Cpath> \u003Cpath> ...`\n   call. One `open` invocation lands all the files in one window with a\n   thumbnail sidebar; separate calls spawn separate windows.\n3. **Caption them** — list one line per shot (what it is) so the user knows\n   what they're flipping through.\n\n### Preview rules\n\n- macOS only (`open`); pass absolute or repo-root-relative paths.\n- Order the files most-important-first; the first one is what Preview shows on open.\n- Review saved PNG\u002FGIF artifacts, not live Chrome. Exact regression capture\n  should run headless by default; only open Chrome when the user explicitly asks\n  for a live browser or when reproducing a headful-only bug.\n- **One set at a time — never pile up windows.** Before opening a new set, close\n  any open ones: `osascript -e 'tell application \"Preview\" to close every window'`.\n- **Close on unattended proceed.** When you opened shots for a non-blocking review\n  and then move on without the user's response, close Preview afterward:\n  `osascript -e 'quit app \"Preview\"'`. An overnight autonomous run must never wake\n  the user to a wall of Preview windows — at most one set open, and none once\n  you've proceeded past the checkpoint.\n","# Screenshot Review\n\nTwo modes for putting eyes on visual artifacts. Pick by audience:\n\n- **Critique** — agent eyes: spawn an unprimed sub-agent to find defects before\n  you accept visual work. Mandatory before declaring a visual bug fixed.\n- **Preview** — user eyes: open the review-worthy shots in a single macOS\n  Preview window so the user can flip through them.\n\nThey compose: run the critique first, then preview the surviving candidate\nshots for the user's sign-off. This skill is for visual defects and human\nreview, not pixel metrics; pair with `compare-screenshots` when you also need\nnumbers.\n\n## Critique mode\n\nUse an unprimed sub-agent as a second set of eyes before accepting visual work.\n\n### Workflow\n\n1. Capture or locate the exact PNGs\u002FGIF frames under review.\n2. Create tight 2x-4x crops for every key feature under judgment, plus the full\n   screenshot for context. Crop selected units, city\u002Ftown stacks, flags\u002Fpoles,\n   shadows, selection rings, labels\u002Ficons, roads, terrain features, water, and\n   any artifact-prone area. If the complaint is about \"too faint\", \"wrong\n   order\", or \"not in perspective\", the crop is mandatory.\n3. Spawn one fresh explorer with `fork_context: false`; pass only the full\n   images, the crops, and a short neutral task. Do not include the main thread\n   history, implementation details, or expected answer.\n4. Ask for concrete visible defects with confidence levels. Name likely risk\n   categories: unit\u002Fprop depth ordering, layering, shadows, selection-marker\n   contrast, ground-plane perspective, flag\u002Fpole attachment, label style and\n   icon readability, blur, scale, lighting, artifacts, missing models, terrain\n   feature readability, roads, water, and overall scan readability.\n5. Compare the sub-agent's critique against your own inspection. Treat overlap\n   as high-priority evidence. Treat novel high-confidence findings as bugs to\n   inspect, not as taste notes to dismiss.\n6. Record actionable findings in the spec, visual report, or next task plan\n   before claiming the screenshot is accepted.\n\n### Sub-Agent Prompt\n\nUse this shape, replacing the bracketed surface and attaching local images:\n\n```text\nFresh visual critique task. You have no project backstory and should only\ninspect the supplied screenshots and crops. First inspect the full screenshot\nfor context, then inspect each crop at zoomed scale. Look for concrete\nvisual\u002Flayout defects in [surface], especially unit\u002Fprop depth ordering,\nlayering, shadows, selection-marker contrast, ground-plane perspective,\nflag\u002Fpole attachment, label style\u002Ficons, blur, scale, lighting, artifacts,\nmissing models, terrain feature readability, roads, water, and scan\nreadability. Do not assume these are correct. Return a concise list of issues\nyou can see, with confidence and whether the issue is visible in the full image,\nthe crop, or both.\n```\n\nSpawn config:\n\n- `agent_type`: `explorer`\n- `fork_context`: `false`\n- attach screenshots as `local_image` items\n- omit model overrides unless the user explicitly requests one\n\n### Critique rules\n\n- **Mandatory before \"fixed\":** never declare a user-reported visual bug fixed\n  on your own inspection — your eyes are primed by the fix you just made. Run\n  the unprimed critique on the candidate shot first; \"mild residue\" you are\n  tempted to wave through is exactly what it exists to catch. (Recorded\n  failure: a \"fixed\" sky that an unprimed agent identified as the terrain\n  mesh's underside filling the entire sky region.)\n- **Reproduce the reporter's framing.** When the user supplied a screenshot,\n  the critique must include a capture at that framing (same camera\u002Fzoom\u002Fspot,\n  or as close as reproducible) — a defect that lives at their framing can be\n  invisible at yours. Your chosen probe framing is a supplement, never the\n  substitute.\n- **Prove the change is real before critiquing it.** Byte\u002Fpixel-diff the\n  candidate against the pre-change baseline first: a critique of an unchanged\n  image \"verifies\" a no-op. (Recorded failure: a palette pass that never\n  reached the production render path — before\u002Fafter were byte-identical and\n  only the diff caught it.)\n- Never tell the sub-agent the defect you expect it to find.\n- Use the current candidate screenshot, not a stale report or baseline image.\n- Do not rely on full-page report scale for small visual features. Attach\n  crops around the exact features a player would read: selected army\u002Fcity,\n  label\u002Ficon clusters, flags, shadows, ring edges, road crossings, terrain\n  feature patches, water labels, and suspicious debug\u002Fartifact regions.\n- If the sub-agent says a crop reveals an issue that is weak or invisible in\n  the full shot, treat it as a real usability defect when the player can zoom\n  to that scale in-game.\n- For animation, attach a short set of deterministic still frames first; GIFs\n  are useful for human review, but still frames make specific defects easier to\n  name.\n- A passing sub-agent critique does not replace direct inspection by the main\n  agent or screenshot regression gates.\n- If the sub-agent catches an issue the main agent missed, add that failure mode\n  to the relevant feature plan or visual checklist immediately.\n\n## Preview mode\n\nPut the review-worthy images in front of the user in a single Preview window\nthey can flip through.\n\n### Workflow\n\n1. **Curate the set** — the images that answer \"is this right?\": the\n   just-changed or just-blessed baselines, the integrated result, and the\n   reference image when there's something to compare against. Don't dump every\n   PNG in the folder; pick the smallest set that lets the user judge the\n   change. If the user named specific shots, open exactly those.\n2. **One window** — open them with a single `open -a Preview \u003Cpath> \u003Cpath> ...`\n   call. One `open` invocation lands all the files in one window with a\n   thumbnail sidebar; separate calls spawn separate windows.\n3. **Caption them** — list one line per shot (what it is) so the user knows\n   what they're flipping through.\n\n### Preview rules\n\n- macOS only (`open`); pass absolute or repo-root-relative paths.\n- Order the files most-important-first; the first one is what Preview shows on open.\n- Review saved PNG\u002FGIF artifacts, not live Chrome. Exact regression capture\n  should run headless by default; only open Chrome when the user explicitly asks\n  for a live browser or when reproducing a headful-only bug.\n- **One set at a time — never pile up windows.** Before opening a new set, close\n  any open ones: `osascript -e 'tell application \"Preview\" to close every window'`.\n- **Close on unattended proceed.** When you opened shots for a non-blocking review\n  and then move on without the user's response, close Preview afterward:\n  `osascript -e 'quit app \"Preview\"'`. An overnight autonomous run must never wake\n  the user to a wall of Preview windows — at most one set open, and none once\n  you've proceeded past the checkpoint.\n",{"name":5,"description":7},"main","e24b64bdac040ad169ed16b14853eff1b20cd726","skills_archive\u002Fscreenshot-review",0,"https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates","https:\u002F\u002Fgithub.com\u002FBumpyClock","2026-11-14 03:12:06",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add BumpyClock\u002Fagent-templates --skill screenshot-review",{"tool":38,"command":39},"bunx","bunx skills add BumpyClock\u002Fagent-templates --skill screenshot-review",{"tool":41,"command":42},"pnpm","pnpm dlx skills add BumpyClock\u002Fagent-templates --skill screenshot-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fscreenshot-review\u002FSKILL.md",true,[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"agents\u002Fopenai.yaml","REFERENCE","参考文档",278,"https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fscreenshot-review\u002Fagents\u002Fopenai.yaml",{"path":53,"resourceType":54,"resourceTypeDesc":55,"sizeBytes":56,"sourceHtmlUrl":16},"LICENSE.txt","LICENSE","许可证",1068,[58,65,72,79,86,93],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":20},"01M79FEE6E257MWVEBY43X0NVN","writing-plans","Use when you have a spec or requirements for a multi-step task, before touching code","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fwriting-plans\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":20},"01M79FEB8GTJC6ZHDK67T40R20","workflow-from-chats","Extract preferences from agent chats → skills, rules, workflow docs. For learning preferences, mining feedback, personalizing workflows.","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fworkflow-from-chats\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":20},"01M79FE7TMNGVZZ0M8NV4N0X6N","winui3-csharp-app","WinUI 3 \u002F Windows App SDK desktop apps in C#\u002FXAML. MVVM, bindings, navigation, Mica\u002FAcrylic, windowing, packaging, performance.","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fwinui3-csharp-app\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":20},"01M79FE31HY46BHN67CQGB7MVJ","why","Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available MCPs and queries each evidence category (source control, issue tracker, long-form docs, real-time chat, infrastructure observability, error tracking, product analytics warehouse) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fwhy\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":20},"01M79FE0N7MGWDG8SCD60FWTV5","what-did-i-get-done","Summarize authored commits over a user-specified time period into a concise status update.","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fwhat-did-i-get-done\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":20},"01M79FDYBJG9QNCQC0MMVBDT80","weekly-review","Produce a short weekly synthesis of authored commits with highlights by bugfix, tech debt, and net-new work.","https:\u002F\u002Fgithub.com\u002FBumpyClock\u002Fagent-templates\u002Fblob\u002Fmain\u002Fskills_archive\u002Fweekly-review\u002FSKILL.md",[],[]]