Conciseness

The body is lean and assumes Claude's competence: no explanation of what Lambda, API Gateway, SAM, or CDK are — just decision guidance ('HTTP API... lower cost and latency; the default', 'REST API... choose only if you need request/response transformations'), terse defaults, complete templates, and deploy commands. Every section earns its place; the runtime version numbers (python3.12, NODEJS_20_X) are embedded in executable config, not temporal advice.

Actionability

Both paths are fully executable and copy-paste ready: a complete `template.yaml` with `sam build && sam deploy --guided`, a complete CDK TypeScript stack with `cdk deploy`, and a concrete verification curl command. The common case (a GET /items HTTP API) is covered end to end, matching the score-5 anchor.

Workflow Clarity

The sequence is clear — decide the shape, apply defaults, pick option A or B, deploy, then a Verify section with three explicit checkpoints ('Call the output URL', 'Confirm structured logs and no permission errors in CloudWatch Logs', 'Check the function's IAM role'). Not score 5 because there is no error-recovery feedback loop: nothing tells Claude what to do when verification fails (e.g., fix a permission error in the role or re-tune the timeout); not score 3 because the checkpoints are explicit, not merely implied.

Progressive Disclosure

Good structure with clear sections (decide shape, defaults, two options, verify) and well-signaled one-level-deep cross-skill pointers ('load the `choose-a-database` skill', 'Follow the `working-with-aws` guardrails'). Not score 5 because the ~100-line body carries two full IaC templates inline with no references/ bundle — one complete quick-start path plus a reference file for the alternate stack (or advanced configurations) would match the 'clear overview pointing to detailed materials' anchor more closely; not score 3 because nothing is buried and no content is misplaced at this scale.

Specificity

Multiple concrete actions are named — 'Build and deploy a serverless HTTP API on AWS using AWS Lambda behind Amazon API Gateway', 'connect a Lambda function to API Gateway', and an 'infrastructure-as-code walkthrough (AWS SAM and AWS CDK) with least-privilege IAM, structured logging, sensible timeout and memory settings, and pay-per-request defaults'. Coverage spans actions, tooling, and practices; not score 4 because there are no gaps — the what, the tools, and the quality practices are all explicit and concrete.

Completeness

It clearly answers what ('Build and deploy a serverless HTTP API... Provides an infrastructure-as-code walkthrough (AWS SAM and AWS CDK)...') and when ('Use when the user asks to create a REST or HTTP API, connect a Lambda function to API Gateway, deploy a serverless endpoint or backend, or scaffold a serverless application') with concrete trigger phrases, exactly matching the score-5 anchor pattern.

Trigger Term Quality

The 'Use when' clause enumerates natural phrases users would actually say: 'create a REST or HTTP API', 'connect a Lambda function to API Gateway', 'deploy a serverless endpoint or backend', and 'scaffold a serverless application', covering synonyms across REST/HTTP/serverless vocabulary. Not score 4 because the synonym coverage is comprehensive rather than 'a few natural terms missing'.

Distinctiveness Conflict Risk

The niche is distinct — serverless HTTP APIs on AWS specifically via Lambda + API Gateway with SAM/CDK — and triggers like 'connect a Lambda function to API Gateway' are unlikely to fire for unrelated skills. It uses third person throughout with no first/second-person penalty; minimal overlap risk only with closely adjacent AWS deployment skills.