Conciseness

The body is lean throughout: "Lessons are the team's memory. A lesson is a file in .opencastle/lessons/, committed with the change you made" and "Nobody sorts memories afterwards" convey only project-specific facts Claude could not know. Tables carry dense decision information ("Still true? | Run"), and no section explains concepts Claude already knows. Anti-drift check: no "minor instances of over-explanation" that would justify a 4.

Actionability

Every workflow ships a copy-paste-ready command: the full "npx opencastle lesson --title ... --problem ... --why ... --cite src/lib/payments.ts:42" example, "rg -i \"CRON_SECRET\" .opencastle/lessons/ || true", "npx opencastle lesson verify <id> --cite <new/path.ts>", "npx opencastle lesson archive <id> --into <file>", and "npx opencastle sync". The main record-lesson case is a complete filled example; templates use appropriate placeholders. Matches the fully-executable top anchor.

Workflow Clarity

Sequences are explicit with validation and feedback loops: the CLI "prints the new id, and refuses a lesson holding a credential or citing a file that does not exist" (error recovery on record); the "When cited code has changed" section is a doctor-triggered verify/move/archive feedback loop; "Resolve two lessons that contradict each other before you merge either" is a checkpoint before merging; and "doctor warns when the index passes about 2,000 tokens" gates graduation. No destructive or batch operation lacks validation, so no cap applies.

Progressive Disclosure

Structure is good: well-headed sections with the bulk of detail appropriately in the body, and exactly one clearly signaled one-level reference ("The categories, and when each applies: [LESSON-CATEGORIES.md](LESSON-CATEGORIES.md)"). Not a 5 because the referenced LESSON-CATEGORIES.md is not present in the skill's bundle (no references/ directory exists), so the link cannot be verified as a real navigation target — a minor organization gap per the 4 anchor.

Specificity

The description lists several concrete actions — "a lesson in .opencastle/lessons/", "how lessons are cited, re-verified and graduated" — which would merit a 4, but it is written with repeated second-person voice ("your memory", "what you learned", "Use when you confirm", "when you touch"), and the rubric's guideline mandates reducing specificity by 1 for second-person voice. Not a 2 because the domain and actions are named concretely, not generically.

Completeness

It explicitly answers what ("Keeps the team's memory: where what you learned goes ... and how lessons are cited, re-verified and graduated") and when, with a long "Use when ..." clause enumerating concrete trigger situations. Both halves are explicit and concrete, matching the top anchor.

Trigger Term Quality

Natural trigger phrases are present: "a fix after a failure", "a convention", "a decision", "a quirk", "a correction about the code", "when doctor reports a lesson's cited code changed", "when the lessons index grows long". Not a 5 because common synonyms such as "lesson learned", "remember this", or "postmortem" are missing; better than a 3 since the trigger vocabulary goes well beyond generic keywords.

Distinctiveness Conflict Risk

The niche is clear — the .opencastle lessons system with distinct triggers like "doctor reports a lesson's cited code changed" and "the lessons index grows long". Minor overlap remains: "when you work out a procedure no skill covers" could also fire a general skill-authoring skill, so it is not the minimal-conflict top anchor.