Conciseness

The body is dense and almost every line is load-bearing (script purposes, chroma contract hex values, craft rules, install commands), assuming Claude's competence with no concept tutorials. It sits at the 4 anchor ('Efficient; minor instances of over-explanation that could be trimmed') rather than 5 because the FREE/no-paid-calls and create-image-fal gating points are stated three times across the intro, Inputs, and Requires sections, and the deterministic/no-AI-render point is repeated twice.

Actionability

Concrete and mostly executable: named scripts with roles ('One `--config`, one MP4'), a real install command ('pip install playwright && playwright install chromium'), a verified config.example.json with the exact field shape, and a fully specified Playwright-unavailable fallback with concrete filenames (chrome.html → chrome-green.png via chrome-devtools MCP). It stops short of 5 because no literal example invocation of one_shot.py or compose_carousel.py is given, so the first command line must be inferred from the scripts themselves.

Workflow Clarity

The pipeline sequence is legible (build_card → headless-Chrome screenshot → compose_carousel, with one_shot as glue, plus an ordered fallback path) and validation exists: the chroma-bleed recovery loop ('Swap the image or tighten the window_mask thresholds if it bleeds') and the final 'watch' QC step. It fits 4 ('most checkpoints present; minor validation gaps') rather than 5 because the happy-path order is implied by the scripts catalog rather than given as an explicit numbered run sequence, and there is no stated check on intermediate outputs (e.g., the green/magenta PNGs) before the encode step.

Progressive Disclosure

The 65-line body is an overview that appropriately defers detail to the four real scripts/, each referenced by exact path with a one-line purpose — verified to all exist, including config.example.json, which matches the body's claims (placeholder /abs/path/... paths, derived duration). References are one level deep, well-signaled under a 'Scripts (free)' section, and navigation is easy, matching the 5 anchor.

Specificity

The description enumerates multiple concrete, specific actions: 'an HTML card (real DOM text) rendered to PNG via headless Chrome, chroma-keyed, its magenta window refilled per-product in PIL, then animated + audio-synthed with FFmpeg' — comprehensive coverage of the pipeline with named tools. It clearly matches the anchor 'Lists multiple specific concrete actions; comprehensive coverage' and is not the level below (only several actions with minor gaps); nothing above 5 exists on the scale.

Completeness

Both halves are present: a very detailed 'what' and an explicit trigger clause 'Use for the airdrop-notification-carousel format.' This fits the 4 anchor ('both what and when; when could be more explicit or specific') rather than 5 because the when-clause is a single narrow phrase naming the internal format, with no richer 'use when the user asks for…' phrasing covering the situations that would invoke it.

Trigger Term Quality

Good natural-keyword coverage — 'viral iOS "AirDrop" notification-carousel video ad', 'product photos', 'carousel', 'Accept tap', 'range/lineup payoff' are phrases a user requesting this ad format would plausibly say. It falls short of 5 because common user-side variations like 'share sheet', 'product showcase', or 'notification-style ad' are absent, and some space is spent on internal jargon ('chroma-keyed', 'audio-synthed') users would not say.

Distinctiveness Conflict Risk

The description carves out a clear niche — deterministic iOS AirDrop share-sheet carousel video ads with chroma-keyed real product photos — with essentially no overlap risk against generic video, image, or document skills. It matches the 5 anchor ('Clear niche with distinct triggers; minimal conflict risk'); the 4 anchor's 'minor overlap risk with closely related skills' does not apply.