Conciseness

The body is dense and information-rich, assumes Claude's knowledge of Convex, and avoids explaining basic concepts; every section (Workflow, Rules) earns its tokens with no padding.

Actionability

Concrete guidance throughout: named MCP tools (status, insights, tables, functionSpec), specific code patterns (.collect()/.filter(), .withIndex, .take(n), .paginate, @convex-dev/sharded-counter), and explicit finding schema fields; as an instruction-only skill the absence of a code block is not penalized because the guidance is specific and executable.

Workflow Clarity

A clear numbered GUARD→GATHER→ROOT-CAUSE→EMIT→REPORT→Scope sequence with explicit validation checkpoints (deploy-guard gate, "if it returns nothing… fall back… do NOT invent findings") and a feedback loop ("re-run insights after traffic to verify the trend").

Progressive Disclosure

Under 50 lines, single-purpose, and well-organized into ## Workflow and ## Rules sections with no need for external bundle references; per the rubric's simple-skill exception this earns a 5 from clean sectioning alone, and the one-level reference to specs/finding.schema.json is clearly signaled.

Specificity

"Read the Convex deployment's 72h insights", "root-cause each event in code", and "report evidence-backed perf/cost findings with fixes" list multiple concrete, distinct actions with comprehensive coverage for the skill's scope; not a 4 because coverage is not merely several actions with minor gaps.

Completeness

The "what" is clear and concrete, but there is no "Use when…" clause or equivalent explicit trigger guidance; per the rubric guideline a missing "when" caps completeness at 3, and not a 4 because the "when" is neither explicit nor specific.

Trigger Term Quality

Domain keywords like "Convex deployment", "read limits", "OCC contention", and "perf/cost findings" are natural for the target audience, but common user phrasings (e.g. "my Convex app is slow") and synonyms are missing; not a 5 because coverage is not comprehensive with synonyms, not a 3 because the present terms are genuinely natural rather than generic.

Distinctiveness Conflict Risk

"Convex deployment's 72h insights", "OCC contention", and "read limits" define a narrow, specific niche with distinct triggers and minimal overlap with other skills.