Conciseness

The body is lean and assumes competence: no explanation of what an AST or code graph is, no library tutorials, and every section (when/when-not, path cheat sheet, workflow, anti-patterns) earns its tokens. It matches anchor 5 (every token earns its place) rather than anchor 4, since there is no noticeable padding to trim.

Actionability

Every operation is a copy-paste-ready command with full paths (`cat ~/.deeplake/memory/graph/query/<pattern>`, `impact/<symbol>`, `path/<from>/<to>`) plus a description of expected output, and placeholders are explained concretely ("a digit from a prior find/ (e.g. 3)"). Matches anchor 5 (fully executable, copy-paste ready, covers common cases).

Workflow Clarity

The 5-step numbered workflow clearly maps task shape to command, and anti-patterns supply error-recovery loops ("Try multiple keywords if the first misses", "prefer the live source for that file"). It sits between anchors 4 and 5: the sequence is clear and mostly checkpointed, but freshness/staleness validation lives in the anti-patterns section rather than as explicit steps in the workflow itself. Since this is a read-only query skill, the destructive/batch validation cap does not apply.

Progressive Disclosure

No bundle files (references/, scripts/, assets/) exist, and none are needed: the ~95-line body is appropriately self-contained, with the path cheat sheet as core overview content that belongs inline, clear section headers, and easy navigation. This matches anchor 5's clean-organization standard rather than anchor 4, which requires some misplacement or organization gap.

Specificity

"Query the local code graph (functions, classes, calls, imports)" names the domain plus one concrete action (query), but does not enumerate the several specific operations the skill actually supports (find, show, impact, path, layers). This matches anchor 3 (domain and 1-2 concrete actions, not comprehensive) rather than anchor 4, which requires several listed actions.

Completeness

It explicitly answers both: what ("Query the local code graph… an AST-derived map of the repo, queried as files") and when ("Use when the user asks structural questions about the codebase") with concrete trigger phrases. This is a direct match for anchor 5.

Trigger Term Quality

The description quotes natural user phrasings verbatim — "what calls X?", "what does Y import?", "where is Z defined?", "what is the architecture / which subsystems exist?" — giving good keyword coverage, but common variants like "usages", "depends on", or "find callers" are missing. Anchor 4 (good coverage, a few natural terms missing) fits better than anchor 5's comprehensive synonym coverage.

Distinctiveness Conflict Risk

A clear niche — structural/relational questions about code via the Deeplake graph mount — with distinct trigger phrasing that would not fire for unrelated skills (writing code, running tests, general file reading). Matches anchor 5 (clear niche with distinct triggers, minimal conflict risk).