Conciseness

The body is dense and operational throughout — phase tables, category tables, exact bash, and a maintainer section explicitly marked "execution does not read this" — with zero explanation of concepts Claude already knows. It falls short of anchor 5 ("every token earns its place") due to redundancy: the opening routing blockquote ("A longer, multi-scene, or narrated treatment → /general-video; a narrated video of a website → /website-to-video...") nearly duplicates the frontmatter description that loads alongside it, the phase table overlaps the Flow steps it summarizes, and the ~20-line maintainer "Design notes" section spends always-loaded tokens on non-execution content.

Actionability

Copy-paste-ready commands for every Bash phase — init ("npx hyperframes init \"$PROJECT_DIR\" --non-interactive --example=blank"), render ("npx hyperframes render . --skill=motion-graphics -q draft -o ./renders/video.mp4"), verify ("npx hyperframes lint . && npx hyperframes inspect ."), preview — plus a one-line validation ("[ -s \"$PROJECT_DIR/shot-plan.json\" ] && echo ok || echo missing"), an enforced subshell convention ("every Bash command... is a (cd \"$PROJECT_DIR\" && ...) subshell"), and exact subagent prompt composition. The single illustrative command in Step 2 is explicitly flagged ("# illustrative — see phases/source/guide.md"), which the guidelines accept as justified flexibility. Common cases (asset-free vs. sourced, mp4 vs. overlay) are each covered.

Workflow Clarity

Steps 0-6 are clearly sequenced with a conditional branch ("If asset_needs is empty, skip to Step 3"), explicit validation checkpoints at each handoff (the shot-plan existence check, lint/inspect with "exit 0 → done"), an error-recovery feedback loop ("On lint/inspect errors, dispatch the repair subagent... one in-place fix pass + re-render"), a graceful-degradation path for missing providers, and a resume table keyed on artifact state ("no shot-plan.json → Step 1", "renders/video.mp4 exists → Report + stop"). This matches anchor 5's feedback loops and checkpoints.

Progressive Disclosure

The body's structure and signaling are strong — the phase table's "Detailed flow" column, "See phases/source/guide.md", and per-category "categories/<id>/module.md" pointers are one level deep and clearly labeled, and the three files under references/ (motion-vocabulary.md, shot-plan-ir.md, builder-contract.md) do exist and match their citations. However, scored against the actual bundle: the majority of referenced detail files — agents/director.md, agents/builder.md, agents/finalize.md, phases/source/guide.md, phases/source/resolve.mjs, categories/<id>/module.md, and catalog-map.md — are absent from the provided bundle, so most navigation pointers dead-end. That is more than the "minor organization gaps" of anchor 4 but well above the inlining/buried-reference problems of anchor 2, leaving anchor 3 (structure present, organization incomplete).

Specificity

The description enumerates a comprehensive set of concrete deliverables — "kinetic typography, stat or number count-up, chart/data-viz hit, logo sting, brand lockup, lower-third, callout, social overlay, animated headline/tweet/news item, motion poster" — plus a concrete output capability ("Can render to MP4 or transparent overlay") and explicit duration boundaries. It falls short of anchor 5 because the action verbs are largely implicit: it never says it creates/animates these artifacts, instead relying on noun-phrase enumeration ("Use when the user wants a short, design-led motion graphic..."); anchor-5 examples state multiple explicit actions ("Extract text... fill forms, merge documents"). It is well above anchor 3, which expects only 1-2 concrete actions with limited coverage.

Completeness

Both halves are explicit with concrete trigger phrases: the "when" is "Use when the user wants a short, design-led motion graphic where motion is the message: [12 concrete trigger types]", and the "what" is the deliverable set plus "Can render to MP4 or transparent overlay". It also adds negative triggers ("Not for longer, multi-scene, narrated, or brand-reel pieces (use general-video)...") that sharpen the when. This matches the anchor-5 example's structure (explicit what + explicit 'Use when' with concrete triggers); anchor 4 would require the when to be merely present but less specific, which is not the case.

Trigger Term Quality

Natural user phrasings are comprehensively covered with synonyms: "kinetic typography", "stat or number count-up", "logo sting", "lower-third", "social overlay", "animated headline/tweet/news item", "motion poster", "transparent overlay", "MP4". These are exactly the words a user asking for this kind of piece would say. It does not fit anchor 4 ("a few natural terms missing") because both the plain-language variants ("number count-up") and the designer vocabulary ("brand lockup", "logo sting") are present.

Distinctiveness Conflict Risk

The description carves out a clear niche (short, unnarrated, motion-first, "Usually under 10s and up to ~30s, with no narration arc, voice-over, or live-action subject") and explicitly disambiguates against six sibling skills by name — "(use general-video)", "(website-to-video)", "(faceless-explainer)", "(product-launch-video)", "(pr-to-video)", "(embedded-captions)" — plus a tie-breaker ("When unsure... see /hyperframes"). Minimal conflict risk with distinct triggers, matching anchor 5.