Content
72%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 reference skill: concrete, specific anti-patterns with exact identifiers and a strong reference-loading strategy that keeps the main file as an overview. Weaker on workflow clarity (no sequenced procedural workflow with checkpoints) and slightly token-heavy due to the inlined dual-platform MCP config.
Suggestions
Move the dual-platform AWS Pricing MCP server JSON config (lines 67-108) into a reference file or collapse to a single example with a one-line Windows variant note, reducing ~40 lines of token weight from the main SKILL.md.
Add a short sequenced workflow with an explicit validation checkpoint for destructive operations (e.g., 1. plan in PR, 2. review plan output, 3. apply only after approval, 4. verify no state corruption) so the scattered apply/destroy safety guidance becomes a clear procedure.
Include one or two minimal positive Terraform code snippets showing the correct pattern (e.g., a scoped Lambda log-group role, an ephemeral secretsmanager resource) alongside the 'Don't' statements to make guidance copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean and assumes Claude's competence — no padded explanations of what Terraform or IAM is — but the ~40 lines of dual-platform MCP server JSON config (lines 67-108) inlined in the main file is token-heavy for a SKILL.md overview and could be trimmed or moved to a reference. Not a 3 because there is no unnecessary concept explanation; not a 5 because the inlined config block is over-explanation that could be tightened. | 4 / 5 |
Actionability | Highly concrete guidance — exact Checkov IDs ('CKV_AWS_28/119', 'CKV_AWS_258', 'CKV_AWS_76/CKV2_AWS_29'), exact IAM actions ('sqs:ReceiveMessage, sqs:DeleteMessage, sqs:GetQueueAttributes'), specific attributes ('authorization_type = "NONE"', 'source_arn'), and copy-paste-ready MCP JSON. Not a 5 because the Terraform guidance is actionable instruction ('Don't do X, do Y') rather than positive copy-paste code examples showing correct patterns — a minor gap consistent with the 'mostly executable' anchor. | 4 / 5 |
Workflow Clarity | This is a reference/anti-patterns skill rather than a procedural one, so there is no sequenced multi-step workflow with explicit validation checkpoints; safety guidance for the destructive 'terraform apply' operation does appear as anti-patterns ('require plan review in PR', 'Don't use cancel-in-progress: true on apply/destroy jobs — cancelling mid-apply corrupts state'), but it is scattered rather than presented as a clear sequence. The reference-loading decision table is clear, but checkpoints are implicit, matching the anchor-3 description; not a 4 because no clear sequenced workflow is present. | 3 / 5 |
Progressive Disclosure | Clear overview body with well-signaled one-level-deep references — all three referenced files (terraform-structure.md, security-iam.md, cicd-patterns.md) exist and are pointed to via both a Quick Reference topic→file table and an explicit 'Reference Loading Strategy' decision table with line counts. The 1300 lines of detail are appropriately split out of the main file, and navigation is easy. | 5 / 5 |
Total | 16 / 20 Passed |