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 strong operational skill with excellent workflow validation, destructive-operation safeguards, and a clean reference structure. Its main weakness is token efficiency: the async waiter/polling boilerplate and selection-rules doc links are duplicated across every action section despite an existing operations reference that centralizes them, and no example selection-rules JSON is provided.
Suggestions
State the waiter/fallback-polling pattern once (e.g., in a short 'Async operations' section) and reference schema-conversion-operations.md for the per-operation waiter and describe-command mapping, removing the verbatim repetition in Auto Import, Convert Database, Assess Database, Work with Tree, and Convert Statement.
Include one concrete example selection-rules JSON (e.g., an 'include' rule for all schemas and an 'explicit' rule for a single table) inline so the most common build step does not depend entirely on fetching external docs.
Cite the selection-rules documentation link once in a shared location instead of repeating the full URL six times across sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The waiter-plus-fallback-polling block ('aws dms wait metadata-model-...' with 'fall back to polling describe-... every 30 seconds') is repeated verbatim across Auto Import, Convert Database, Assess Database, Work with Tree, and Convert Statement, even though references/schema-conversion-operations.md already documents all waiter and polling mappings; the selection-rules documentation link is repeated six times. It could be tightened by stating the pattern once and referencing the operations table, though there is no padding explaining concepts Claude already knows. | 3 / 5 |
Actionability | Commands are concrete and near copy-paste ready — exact CLI calls, waiter invocations with filters, --properties JSON for statement conversion, S3 key extraction. The gap is that selection rules, the most frequently required input, are never shown as an example JSON and remain as '<json>' placeholders deferred to external docs. | 4 / 5 |
Workflow Clarity | Every async operation has an explicit validation checkpoint (DMS waiter with terminal statuses SUCCESS/FAILED and a polling fallback), failures route to a troubleshooting reference forming a feedback loop, and destructive cleanup is gated by customer confirmation with mandatory dependency-ordered deletion. Sequences are numbered and unambiguous throughout. | 5 / 5 |
Progressive Disclosure | The body is a well-organized hub: each action workflow links to real one-level-deep reference files (setup-wizard.md, action-items.md, troubleshooting.md, cancel-operations.md, schema-conversion-operations.md — all verified to exist), and the 822-line setup wizard is correctly externalized. Navigation is clear with anchors and point-of-use links. | 5 / 5 |
Total | 17 / 20 Passed |