Content
92%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.
An exemplary instruction-only guard: every step is executable with exact flags, commands, and file paths, and the risky operations are fenced with consent, ambiguity, and disagreement checkpoints. The only real criticism is redundancy between the Workflow steps and the Rules bullets that restate them.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and directive with no background tutorials, but five of the six Rules bullets restate Workflow steps nearly verbatim (e.g. step 6 vs "A deploy that seemed to do nothing means the WRONG deployment changed") and the opening motivational sentence could be trimmed. | 4 / 5 |
Actionability | Fully actionable for an instruction-only skill: exact env vars and files (CONVEX_DEPLOYMENT in .env.local, convex.json, CONVEX_DEPLOY_KEY), literal commands (npx convex deploy, npx convex run --prod, npx convex env list), exact MCP flags, and a copy-paste announce format ('target: dev (joyful-capybara-123, personal dev)'). | 5 / 5 |
Workflow Clarity | Seven explicitly sequenced steps with validation checkpoints ("If two sources disagree, resolve before proceeding", "Ambiguity = stop... never guess") and a feedback loop for the failure case (a no-op deploy means re-run step 1, not re-deploy harder), so the destructive-operation cap does not apply. | 5 / 5 |
Progressive Disclosure | Under 50 lines, no bundle files exist to reference, and the content is well-organized into Workflow and Rules sections, satisfying the simple-skill exception for progressive disclosure. | 5 / 5 |
Total | 19 / 20 Passed |