Select an evidence-backed platform delivery profile, preserve organic and ads scope boundaries, and produce safe-zone QA input with an explicit human preview gate when evidence is unknown or stale.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Select the registered platform and surface profile before delivery review. Keep profile evidence, freshness, safe-zone geometry, fallback, and human preview separate from the render and from the platform's own preview.
Select the exact platform, surface, and delivery variant under delivery_profiles/platform-safe-zone. Read source URL, retrieved date, published date, owner, evidence status, scope, supersession, and profile hash.
Keep organic and ads profiles separate. Reject an ads profile for an organic surface and reject organic evidence as an ads measurement.
For a verified profile, verify device evidence, viewport/output geometry, pixel density, UI and safe regions, method, confidence, screenshot path, and SHA256. Use the existing visual-status producer, which loads the registered profile and runs its regression runtime, before delivery review:
npx tsx scripts/caption-review.ts visual-status --project <project> --safe-zone-profile <project-relative-profile>
For partial, unavailable, stale, or unknown evidence, keep regions unknown, select the registered fallback, and require a human platform preview. Do not infer or add coordinates.
Treat regression as an artifact check for known anchors, captions, content, and profile overlays. The human platform preview is a separate gate and is required whenever the profile says so.
When a current measurement is authorized, record platform, surface, iOS or Android, device, viewport resolution, pixel density, notch or insets, app version and build, locale, and measurement date. Use a rights-cleared test clip, capture a lossless screenshot, record its SHA256, and record the measurement method and confidence. Check viewer variants separately, isolate ads from organic, and retain unknown status when any required evidence is absent or stale.
Run npm run validate:all-local, npm run typecheck, and npm run verify. Inspect delivery_profiles/platform-safe-zone and the safe-zone regression result (platform-safe-zone-qa/v1) when emitted by the runtime and route receipt; do not invent a separate receipt file. Keep platform preview evidence and renderer output as separate artifacts; this skill does not authorize compositor or Studio UI work.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
MUST USE when source footage or a material folder is provided and material understanding or 03_analysis generation is needed. Trigger not only on explicit analysis requests ('分析して', '素材を見て') but also when the user points at footage and says things like 'この動画で', 'この素材で', or any request that requires understanding the provided media before planning or editing.
The English summary is being prepared. Showing the original description.
Use when selects_candidates.yaml exists and the user asks to design structure, build an edit blueprint, or create 04_plan/edit_blueprint.yaml.
The English summary is being prepared. Showing the original description.
Use when edit_blueprint.yaml and selects_candidates.yaml are ready and the user asks to compile the rough cut, generate timeline.json, or apply a review patch.
The English summary is being prepared. Showing the original description.
MUST USE when the user asks to crop, reframe, trim, or improve composition of still photos, especially selected event photos, album photos, portraits, speakers, panels, workshops, or group photos. Use after photo selection or correction when the task is composition improvement rather than selecting images.
The English summary is being prepared. Showing the original description.
MUST USE when starting a new video project, when the user provides footage and asks to edit/create a video, or when 01_intent/creative_brief.yaml does not yet exist. This skill interviews the user to understand their intent before any editing begins. Never skip this step.
The English summary is being prepared. Showing the original description.
Use after changing prompts, skills, the compiler, or editorial policies — or when the user asks to measure edit quality, run regression evals, check agreement with approved edits, or says '評価して', '一致率を測って', 'eval を回して'. Compares agent output against human-approved golden projects.
The English summary is being prepared. Showing the original description.