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.
A well-structured, highly actionable skill body with an explicit multi-step workflow, validation checkpoints, error-recovery guidance, and clean progressive disclosure into two real reference files. The main weaknesses are repeated constraint statements that pad the token budget and a missing concrete command for retrieving query results/cost in the final step.
Suggestions
Deduplicate constraint statements: 'confirm the output S3 location', 'respect the user's decision to abort', and the workgroup-first rule each appear 2-3 times across the Overview, task sections, and Principles — state each once in the most relevant section.
Add the concrete command for retrieving results and query statistics in step 7 (e.g., 'aws athena get-query-execution / get-query-results --query-execution-id <ID>') so the execute-then-present loop is fully executable.
Trim or merge the Principles section, which largely restates constraints already covered in the Common Tasks sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense and efficient (constraint bullets, tables, exact addressing rules) but several constraints are stated multiple times: 'You MUST confirm the output S3 location before executing' appears in both the Overview and task 6, 'You MUST respect the user's decision to abort' appears twice, and the workgroup-first rule is repeated in task 2 and Principles. This repetition goes beyond the 'minor instances' of the score-4 anchor, fitting 'mostly efficient but could be tightened'. | 3 / 5 |
Actionability | Guidance is largely executable: 'aws sts get-caller-identity', a complete 'aws athena start-query-execution --work-group <WORKGROUP_NAME> --query-string ...' example, catalog-specific addressing rules, a statement classification table, and an error/cause/fix table. The minor gap is that step 7 (Present and Recover) gives no command for fetching results or query statistics (e.g., get-query-results/get-query-execution), keeping it just below the fully-executable score-5 anchor. | 4 / 5 |
Workflow Clarity | Seven clearly sequenced tasks with explicit validation checkpoints (verify credentials first, select a workgroup before submitting, state the resolved asset, classify statements and confirm destructive ones) and feedback loops for error recovery via the troubleshooting table and 'On failure, list available workgroups and let the user choose which to retry with'. Argument routing is explicitly ordered ('Resolve in this order; stop at the first match'), matching the score-5 anchor. | 5 / 5 |
Progressive Disclosure | The body is a clear overview that defers detail to two real, one-level-deep references, each well-signaled at its point of use (workgroup-selection.md in task 2, query-patterns.md in argument routing) and again under Additional Resources. Content is appropriately split with easy navigation, matching the score-5 anchor. | 5 / 5 |
Total | 17 / 20 Passed |