Content
16%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 reads as a persona/résumé document rather than a skill: it catalogs everything Claude already knows about Terraform while providing no executable guidance, commands, or examples. The small salvageable parts are the 'Use this skill when'/'Do not use' sections, the four-step instruction sketch, and the safety notes; the rest should be deleted or moved into reference files.
Suggestions
Delete the 'Capabilities', 'Knowledge Base', 'Behavioral Traits', 'Purpose', and 'Example Interactions' catalogs — Claude already knows Terraform's feature set — and keep only the ~30 lines of trigger conditions, instructions, and safety rules.
Add concrete, executable guidance to the 'Instructions' section: e.g. a minimal remote backend configuration snippet (S3 + DynamoDB locking), the terraform plan/apply review workflow with an explicit 'review plan before apply' checkpoint, and the OpenTofu migration command sequence.
Move any genuinely needed detail (backend-specific configs, migration compatibility matrices, CI/CD pipeline examples) into one-level-deep reference files under references/ and link to them from a concise overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly 130 of 155 lines are catalogs enumerating concepts Claude already knows — ten 'Capabilities' sections listing Terraform features, backend names (S3, Consul, etcd), testing frameworks, and compliance frameworks — plus a 'Purpose' section duplicating the frontmatter description and a 'Behavioral Traits'/'Knowledge Base' persona dump. Severely verbose and heavily padded with no information Claude doesn't have. | 1 / 5 |
Actionability | There is no code, command, configuration snippet, or concrete example anywhere in the body. The 'Instructions' are four abstract one-liners ('Define environments, providers, and security constraints') and 'Response Approach' is nine abstract steps — it describes rather than instructs, matching anchor 1. | 1 / 5 |
Workflow Clarity | The 'Instructions' section presents a coherent design sequence (define constraints → design modules/backend → implement plan/apply workflows → validate) and validation is at least named ('Validate drift, costs, and rollback strategies'; Safety: 'Always review plans before applying changes'), so it sits above anchor 2's 'validation absent'. But steps are one-line abstractions with no executable checkpoints or fix-and-retry feedback loops, and destructive apply operations cap the score at 3 regardless. | 3 / 5 |
Progressive Disclosure | No bundle files exist and nothing is split out — the ~130-line capability catalog, which clearly belongs in separate reference files, is inlined wholesale in SKILL.md with no pointers or navigation. Section headers exist, but there is zero disclosure layering, matching anchor 2 ('content that clearly belongs in separate files is inlined'). | 2 / 5 |
Total | 7 / 20 Passed |