Content
85%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 well-structured, highly actionable guard skill with explicit validation and feedback loops for destructive deployment operations. Its main weakness is the Rules section duplicating the Workflow, adding token cost without new information.
Suggestions
Collapse the "Rules" section or fold only the non-duplicated points (e.g., the composition note) back into the Workflow to remove restatement and recover tokens.
Trim the opening motivational sentence ("Deployments are not interchangeable, and most incidents start with..."), which restates context Claude can infer from the workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and assumes Claude knows Convex, but the "Rules" section substantially restates the "Workflow" steps (prod consent, flag splitting, read-only absoluteness, wrong-deploy diagnosis), which is redundant padding that could be tightened. | 3 / 5 |
Actionability | Concrete, executable guidance throughout: specific commands ("npx convex deploy", "npx convex run --prod", "convex env list"), exact MCP flags (--cautiously-allow-production-pii, --dangerously-enable-production-deployments, --disable-tools run,envSet,envRemove), and a copy-ready announce example. | 5 / 5 |
Workflow Clarity | A clearly sequenced 7-step workflow with explicit validation checkpoints (resolve source disagreement in step 1, a diagnose-don't-redeploy feedback loop in step 6, and a stop-on-ambiguity rule in step 7) for destructive deployment operations. | 5 / 5 |
Progressive Disclosure | The skill is under 50 lines, no external references are needed, and the body is well-organized into Workflow and Rules sections — meeting the simple-skill exception for a top score. | 5 / 5 |
Total | 18 / 20 Passed |