Content
77%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.
The body is an operationally rich, well-sequenced guide with real commands, explicit validation gates, and high-value anti-patterns covering failure modes. Its main defects are a repeated ENI-churn explanation, a stub code example, and a progressive-disclosure failure: it cites a nonexistent EXAMPLES.md and never links the bundled sns-testing-guide.md reference.
Suggestions
Fix the broken reference graph: either add the missing EXAMPLES.md walkthrough or point the sentence at the existing references/sns-testing-guide.md, and add a clearly signaled entry for sns-testing-guide.md (e.g. in a references table alongside the scripts).
Deduplicate the Lambda VPC/ENI-churn explanation — keep it once (in the "No interruption" anti-pattern, where it is load-bearing) and drop the near-verbatim copy from Mindset.
Replace the placeholder TypeScript stack skeleton with a minimal runnable example (a concrete resource and a context-driven property toggle) so the design step is copy-paste executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and nearly every section carries non-obvious operational content (validation gates, rollback recovery, DeletionPolicy timing, jq normalization). The main trim opportunities are the Lambda VPC/ENI-churn passage repeated nearly verbatim in both Mindset and the "No interruption" anti-pattern, and a slightly long Mindset preamble — minor over-explanation, not systemic padding. | 4 / 5 |
Actionability | Mostly executable guidance: real `cdk deploy` and `aws cloudformation describe-stack-events --query` commands, a `continue-update-rollback` recovery command, a jq normalization pipeline, and a findings-report template. The TypeScript stack example is a placeholder skeleton ("Add only the resource under test, driven by this.node.tryGetContext(...)") and commands use `<stack-name>` placeholders, which keeps it just short of copy-paste ready. | 4 / 5 |
Workflow Clarity | The four-step workflow (Research → Design → Execute → Document & Decide) has an explicit "Validation gates" block (stop on failed initial deploy, abort and document on unexpected behavior, repeat on ambiguity) plus a decision framework mandating hypothesis-first testing and repetition before trust. Error-recovery paths are also covered by the UPDATE_ROLLBACK_FAILED anti-pattern with a concrete recovery command — this satisfies the destructive-operation validation requirement fully. | 5 / 5 |
Progressive Disclosure | The scripts table cleanly signals both bundled scripts, but the body points to "EXAMPLES.md in this skill directory" for the full walkthrough and that file does not exist, while the actual bundled reference (references/sns-testing-guide.md) is never linked from SKILL.md. Navigation is therefore broken in both directions — a dead reference and an orphaned file — which is more than a minor organization gap. | 3 / 5 |
Total | 16 / 20 Passed |