Content
86%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A lean, highly actionable skill body built around executable commands, exact import paths, and a concrete self-verification feedback loop, with clean section structure and no bundle dependencies.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense and directive with no padding about what Convex or basic programming is; every rule encodes a non-obvious gotcha. Not a 5 because some rationale clauses ('Forcing anonymous on a signed-in user rebinds .env.local and costs them...') and the trailing justification in SELF-VERIFY could be trimmed. | 4 / 5 |
Actionability | Fully executable guidance: real commands ('npx tsc --noEmit', 'npx convex dev --once', 'CONVEX_AGENT_MODE=anonymous npx convex dev --once'), exact import paths ('./_generated/server', './_generated/api'), and concrete API patterns (.withIndex, .paginate, .take(n), v.literal, 'use node'). | 5 / 5 |
Workflow Clarity | A clear 5-step Workflow plus an explicit SELF-VERIFY feedback loop (run tsc, push, fix every error before finishing) and a DEGRADATION RULE for edge cases. Not a 5 because the verification checkpoint lives in Rules rather than being integrated into the numbered workflow sequence, a minor gap. | 4 / 5 |
Progressive Disclosure | Under 50 lines with no external references needed and well-organized into ## Workflow and ## Rules sections; per the simple-skill exception this scores 5 on structure alone, and no bundle files exist to verify. | 5 / 5 |
Total | 18 / 20 Passed |