Conciseness

The body is dense, assumes competence (no basic HTML/GSAP explanations), and earns most of its tokens — but three rules are each stated twice within ~70 lines: the `gsap_css_transform_conflict` rule appears in "Agent pitfalls" and again in "First-pass lint gotchas"; the `.clip` autoAlpha/visibility ban appears three times (Agent pitfalls, lint gotchas, and non-negotiable rules); and root sizing (`100%` not `1920px`) appears in both "Root must be sized" and the lint gotchas list. Sentences like the media-nesting rule carry long multi-clause parentheticals that could be tightened. This is more than the 'minor instances' of anchor 4, though well short of padded concept explanations, so it lands at 3.

Actionability

Mostly executable guidance: named lint rules with their exact fixes (`gsap.fromTo(el, { x: -40 }, { x: 0 })` instead of CSS `transform: translateX(-40px)`), exact CLI commands (`npx hyperframes timeline [--json]`, `npx hyperframes snapshot --at <midpoints>`), and specific attribute contracts (`window.__timelines["<id>"]` keyed to `data-composition-id`). Not a 5 because the complete copy-paste-ready composition skeleton lives in `references/minimal-composition.md` rather than inline — defensible, but the body itself has only one small code snippet.

Workflow Clarity

The build → edit → validate flow is clear, and the final Validation section is an explicit checklist of concrete commands with an ordering gate ("`npx hyperframes render` only after the user approves") plus a trap explanation (a lint error silently disables the layout/contrast audits). It falls short of anchor 5 because no explicit validate → fix → re-run feedback loop is described — error recovery is implied by the lint-rule list rather than sequenced.

Progressive Disclosure

The body is a genuine overview/build guide with a purpose-labeled references table ("Read it to…") covering all 10 files; every referenced path exists in `references/` and is substantive (verified, e.g. `minimal-composition.md` contains a complete runnable skeleton), all references are one level deep, and detail is correctly delegated per topic. Cross-links to sibling skills are clearly signaled. This matches the anchor-5 example structure well.

Specificity

The description enumerates concrete, specific artifacts of the domain ("the `data-*` timing attributes", "`class=\"clip\"`", "tracks, sub-compositions, variables, framework-owned media playback, deterministic-render rules, and validation") beyond the single action "build one renderable project". It sits between anchor 3 (only 1-2 concrete actions) and anchor 5 (multiple distinct actions with comprehensive coverage) — the coverage is comprehensive but reads as a topic inventory rather than a list of distinct action verbs, so it is noticeably above the midpoint but not a full 5.

Completeness

It explicitly answers both questions: what ("The HyperFrames composition contract — build one renderable project") and when ("Use for composition structure, the `data-*` timing attributes, ... and validation. Read before writing composition HTML"). The trigger guidance is explicit and concrete — including the particularly strong pre-task trigger "Read before writing composition HTML" — matching the anchor-5 example structure, whereas a 4 would have the "when" merely present but less explicit.

Trigger Term Quality

Good natural keyword coverage for the domain: "composition", "tracks", "sub-compositions", "media playback", "validation", "composition HTML", "timing attributes" — phrases a user working with HyperFrames would actually say. Not a 5 because it offers no synonyms or file extensions and some terms lean technical ("deterministic-render rules", "framework-owned media playback") rather than colloquial.

Distinctiveness Conflict Risk

The vocabulary is entirely HyperFrames-specific (`data-*` timing attributes, `class="clip"`, sub-compositions, deterministic-render rules), giving it a clear niche with distinct triggers and minimal conflict risk against generic web/HTML skills. Nothing in the description is generic enough to fire for the wrong skill.