Conciseness

The body is dense with operational rules ("Track generation attempts by `stageId + order`", "one page per call, never more") and explains no concepts Claude already knows. A few asides could be trimmed — "the same information a planned outline slot used to carry" and "so there is no need to read each page's source to check `audioId`" — which places it at anchor 4 rather than the every-token-earns-its-place lean of anchor 5.

Actionability

Every step names exact tools and arguments: `create_stage` with `folderId`, `set_roster` with `providerId::voiceId` bindings, `generate_scene` with `title`/`type`/`brief` plus `materialFacts`, `read_stage` at `path:/scenes/<order|id>/actions` with `detail:"source"` following `nextOffset`, and concrete failure codes (`prompt-unavailable`, `invalid-model-output`) with an exact retry-once policy. For an instruction-only skill this is fully executable guidance covering the common cases, matching anchor 5.

Workflow Clarity

The numbered build sequence has explicit validation checkpoints — `list_scenes` to verify every settled page persisted, `generate_tts` reporting generated/skipped/failed counts, and a `read_stage` actions sweep with blank-text and off-topic-narration defect checks — plus feedback loops for error recovery (change the brief without changing the teaching intent, retry once, then leave the gap named) and a done-checklist. This matches anchor 5.

Progressive Disclosure

No bundle files exist and none are needed: this is a single tightly-coupled procedure whose ~100 lines of rules all apply at build time, organized under clear section headers (build sequence, failure handling, narration audio, done criteria). The sibling-skill pointers (`pro-editing`, topic skills) are one level deep and clearly signaled, so per the no-external-references-needed guideline this scores 5.

Specificity

"Covers how the plan is settled in conversation, the order the generation tools are called in, how the classroom roster is written, when narration has to be re-synthesized, and what must be true before a stage is called done" lists five distinct capability areas with only minor coverage gaps. It sits at anchor 4 rather than 5 because the capabilities are phrased as 'how X is done' topics rather than multiple concrete actions.

Completeness

Both questions are explicitly answered: the 'what' is the enumerated coverage of planning, tool order, roster, narration re-synthesis, and done criteria, and the 'when' is concrete trigger guidance — "Load this before generating or rebuilding any stage from scratch, whatever its subject". This matches anchor 5's explicit what-and-when with concrete triggers, not anchor 4 where the 'when' is weaker or less specific.

Trigger Term Quality

Natural terms a user would say are present: "building", "stage", "a single classroom", "generating or rebuilding any stage from scratch", plus domain terms like roster and narration. A few natural synonyms (e.g., lesson, course, deck) are missing, matching anchor 4 rather than the comprehensive synonym coverage of anchor 5.

Distinctiveness Conflict Risk

"it is the floor the topic skills build on, not an alternative to them" and "from scratch" explicitly carve the niche apart from topic skills and from editing, giving mostly distinct triggers with only minor overlap risk against the closely related topic-skill ecosystem. Not anchor 5 because 'stage'/'classroom' triggers still sit adjacent to those sibling skills.