Content
87%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.
Excellent token efficiency and actionability with exemplary progressive disclosure — real, well-organized one-level-deep reference files. The one substantive weakness is workflow clarity around destructive/batch operations: a whole-table `db.delete("person")` and bulk `db.insert` are demonstrated without any caution, error-handling, or verification checkpoint, which the rubric caps at 3 for database skills.
Suggestions
Add an explicit warning next to `db.delete("person")` that it deletes every record in the table, and show how to scope deletes to a specific RecordId or a WHERE clause as a safety checkpoint.
Include a brief verification pattern after bulk operations (e.g., `const count = await db.query<[number]>("SELECT count() FROM person GROUP ALL")`) so batch inserts/updates can be confirmed.
Show basic error handling around the connect/signin sequence (try/catch or a check that signin succeeded before running queries) to add a validation checkpoint to the main workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and code-first: it never explains what SurrealDB is or how libraries work, and every line is either executable guidance ("Always pass user input via bound parameters, never string interpolation") or a tight code example. Every token earns its place. | 5 / 5 |
Actionability | All snippets are copy-paste-ready TypeScript covering the common cases: full connect/use/signin/close flow, all CRUD variants with RecordId, parameterized db.query, and a live-query subscription with kill. This matches the 'fully executable, covers common cases' top anchor. | 5 / 5 |
Workflow Clarity | The connect sequence is explicit and the injection-safety rule is stated, but destructive and batch database operations are shown with no validation, warning, or verification: `await db.delete("person")` deletes an entire table and `db.insert` bulk-inserts with no checkpoint. The rubric explicitly caps workflow clarity at 3 for database operations missing validation/verification steps. | 3 / 5 |
Progressive Disclosure | The SKILL.md is a clear overview with well-signaled, one-level-deep references: in-body pointers ("See [references/embedded.md]", "See [references/live-queries.md]") plus an annotated References section; all three bundle files exist, are substantive, and contain no nested references. | 5 / 5 |
Total | 18 / 20 Passed |