Conciseness

The body is lean and assumes Claude's competence — the one-line framing ("The diff tells you what changed; it does not tell you what breaks") sets scope rather than padding, and heuristics like "follow re-exports one more hop than feels necessary" earn their tokens. No explanations of concepts Claude already knows, matching the lean anchor 5 rather than anchor 4's minor over-explanation.

Actionability

Guidance is concrete for an instruction-only skill: it names a specific tool ("`ast_grep_search` MCP tool"), exact classification criteria, and check types ("its test, its typecheck, its build, or a one-off command"). It stops short of anchor 5's copy-paste-ready bar — no example commands or report template — but exceeds anchor 3, which requires pseudocode-level vagueness or missing key details.

Workflow Clarity

Five clearly sequenced steps with an explicit, mandatory validation checkpoint (step 4: "run the exact check that would catch the break... Report real output") and honest reporting of the unverified remainder in step 5. Not anchor 4, which requires minor validation gaps — there are none; and the destructive/batch cap does not apply because the skill states "The report never edits code."

Progressive Disclosure

The skill is under 50 lines, needs no external references, and is organized into clear sections (When to use / Workflow / Rules) with nothing inlined that belongs in a separate file. Per the rubric's simple-skill guideline this matches anchor 5; no bundle files exist to reference.

Specificity

"Map what a change could break beyond the diff — dependents found by search and AST, classified certain/likely/unlikely — and prove... by running their checks" lists multiple concrete actions covering the entire capability chain (map, classify, verify) with specific tooling detail and no padding. This matches the comprehensive-coverage anchor rather than the 'minor gaps' anchor at 4.

Completeness

Both parts are explicit: the what ("Map... dependents... classified certain/likely/unlikely... prove... by running their checks") and the when ("Before risky refactors or signature changes"). The when-clause is present but narrow — it omits other triggers the skill body itself lists (dependency bumps, migrations, "what could this break?" requests) — fitting anchor 4 rather than anchor 5's concrete trigger-phrase breadth.

Trigger Term Quality

Natural terms like "risky refactors", "signature changes", and "break" are present and would be said by users, but common variations such as "what could this break", "impact", "breaking change", or "downstream" are missing. Good keyword coverage with a few natural terms missing fits anchor 4; it falls short of anchor 5's comprehensive synonym coverage.

Distinctiveness Conflict Risk

"What a change could break beyond the diff" plus "dependents" and "prove by running their checks" carves a clear pre-refactor impact-analysis niche with distinct triggers, minimally overlapping code-review or refactoring skills. This matches anchor 5's clear-nitch description; anchor 4 would require identifiable overlap with a closely related skill, which is not the case.