Content
40%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.
This is a well-coded but bloated monolith: the SDK v3 examples are genuinely actionable, yet the body re-explains AWS fundamentals Claude already knows, offers no task-level workflows or validation for destructive/batch operations, and inlines everything instead of splitting per-service detail into reference files. It functions as an AWS encyclopedia rather than a skill that changes Claude's behavior.
Suggestions
Split the body into one-level-deep reference files (e.g., references/s3.md, references/dynamodb.md, references/cloudformation.md) and keep SKILL.md as a short overview with "When to Use" guidance and per-service pointers.
Delete or compress the AWS-knowledge padding — IAM component definitions, S3 storage-class list, EC2 instance types, and the generic Security/Cost/Reliability best-practice bullet lists — down to only AWS-specific conventions Claude would not infer.
Add validation checkpoints to the destructive/batch paths shown (TerminateInstances, DeleteStack, BatchWrite, DeleteObjects), e.g., DescribeStacks status polling after CreateStack and confirming no errors before terminating instances.
Fix the code defects: replace the v2 "AWS.DynamoDB.Converter.unmarshall" calls with the v3 unmarshall import, merge the duplicate ExpressionAttributeValues object in the Query example, and define the missing "client"/"context" variables.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 1,556-line body extensively re-teaches AWS knowledge Claude already has — IAM users/groups/roles, S3 storage classes, EC2 instance families, credential chains, and ~110 lines of generic Security/Cost/Performance/Reliability best-practice bullets. Anchor 1's pattern ("extensively explains concepts Claude already knows") is largely met, but the genuinely useful SDK v3 code keeps it at anchor 2 rather than 1; time-sensitive details ("Last Updated: October 2025", nodejs20.x runtime) are also not isolated in a deprecated/old-patterns section. | 2 / 5 |
Actionability | Most guidance is concrete, executable SDK v3 code covering common cases (S3 upload/download/pagination, multipart upload with abort-on-error, DynamoDB CRUD/query/batch, Lambda handlers, CloudFormation deploy). Minor gaps keep it from 5: the DynamoDB Streams example uses the v2 API ("AWS.DynamoDB.Converter.unmarshall"), the Query example declares duplicate "ExpressionAttributeValues" keys so the filter values are silently dropped, the multipart-upload snippet references an undefined "client", and the Lambda error handler uses a "context" parameter missing from its function signature. | 4 / 5 |
Workflow Clarity | The body is a per-service reference, not a sequenced workflow — there is no ordered procedure with checkpoints for any task (e.g., deploy a stack, wire Lambda to S3). Destructive and batch operations (TerminateInstancesCommand, DeleteStackCommand, DeleteObjectsCommand, BatchWriteCommand) appear with no validation or verification steps, which caps the score at 3 per the scoring notes, and the overall absence of any task sequence matches anchor 2 ("steps poorly defined; validation absent"). | 2 / 5 |
Progressive Disclosure | There are no bundle files (no references/, scripts/, or assets/), and the entire ~1,500-line encyclopedia — including per-service deep dives like the 160-line CloudFormation template and full DynamoDB API coverage — is inlined in SKILL.md. This matches anchor 2 ("content that clearly belongs in separate files is inlined"); it does not reach 3 because there are no references to signal at all. | 2 / 5 |
Total | 10 / 20 Passed |