Conciseness

The body is ~10 lines of pure commands and operative rules with zero padding — no explanation of what code search is or how indexing works. Lines like "Trust the passage over memory: it is the actual code at that path. Cite file:line for anything you rely on" add guidance Claude could not infer, so every token earns its place.

Actionability

Every entry point gets an exact, executable command line: "xerj search --prefix ref \"<plain words>\"", "xerj def --prefix <ref|trackdub> \"<symbol>\"", "xerj autoindex <folder>", and the MCP "xerj_search" tool. It falls short of 5 only because there is no concrete example query or sample of what a returned passage looks like, leaving 'specific examples cover the common cases' partially unmet.

Workflow Clarity

The trigger is unambiguous ("Use XERJ instead of grep when you need code you cannot see in the open files") and each command is self-contained, but sequencing guidance is thin: nothing states that a repo must be "autoindex"-ed before it can be searched, and there is no guidance on what to do when a search returns nothing. Clear with minor validation/sequencing gaps fits 4 rather than 5.

Progressive Disclosure

This is a single-file skill under 50 lines with no bundle files (no references/, scripts/, or assets/ exist) and no need for external references. Per the rubric's simple-skill guideline, the well-organized trigger-then-commands-then-rule structure earns a 5.

Specificity

"code search over locally indexed repos (xerj search / xerj def)" names the domain plus two concrete actions (search, go-to-definition) with their command names, but omits indexing (autoindex) and the MCP surface. This matches the 'domain and 1-2 concrete actions, not comprehensive' anchor rather than 4's 'several specific actions' since only two capabilities are stated.

Completeness

The 'what' is clear ("Definition-first code search over locally indexed repos") with concrete command names, but there is no 'Use when...' clause or equivalent explicit trigger guidance, which caps completeness at 3 per the judging guidelines. Not 4 because the 'when' is not even weakly implied in the description itself.

Trigger Term Quality

"code search" and "def" are relevant keywords, but the natural phrases a user would actually say — "go to definition", "find symbol", "find usages", "grep" — are absent, and "Definition-first" is jargon rather than a user phrase. This fits 'some relevant keywords but missing common variations or synonyms', not 4's 'good keyword coverage'.

Distinctiveness Conflict Risk

The unique tool name "xerj" and the qualifier "locally indexed repos" carve out a distinct niche with minimal confusion risk. Not 5 because the trigger phrasing "code search" is generic and overlaps with other code-search/grep-adjacent skills; it is 'mostly distinct; minor overlap risk with closely related skills'.