Content
81%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.
An exceptionally actionable, well-sequenced operating manual with real validation checkpoints and feedback loops for database and migration work. Its weaknesses are redundancy (a duplicated comparison table and warnings repeated 3+ times) and inlining detail that belongs in the existing reference files, which hurts token efficiency.
Suggestions
Remove the second, partly-Chinese copy of the NoSQL-vs-PG method table in 'JS SDK v3 PostgreSQL Patterns' (including its 'Golden rule' paragraph) — the table in Required Flow plus the 'Common query helpers' list already covers it.
State each repeated warning once — 'auth.uid() returns text not uuid', 'use auth.getSession() not auth.getUser()', and 'bucket must exist before upload' each appear 3+ times — and point to the relevant reference file (auth-and-rls.md, storage-pg.md) instead of repeating.
Move the storage failure-mode cheat sheet and the extended migration-action details (fetchMigration/repairMigration, out-of-order versions) into references/storage-pg.md and a migration-focused reference, keeping SKILL.md to the core flow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense, non-obvious platform guidance, but the NoSQL-vs-PG method table appears twice (in "Required Flow" and again in "JS SDK v3 PostgreSQL Patterns", the second copy partly in Chinese), and warnings like "auth.uid() returns text, not uuid" and "only use auth.getSession(), never auth.getUser()" each repeat 3+ times across sections. This is more than anchor 4's "minor instances" of over-explanation, but the bulk still earns its place, so it does not fall to anchor 2's padded level. | 3 / 5 |
Actionability | Provides copy-paste-ready TS snippets (SDK init, CRUD shapes, count/pagination/upsert/RPC), exact MCP call signatures with parameters (e.g. managePgDatabase(action=applyMigration, migrationName=..., confirm=true)), a full SQL owner-table template, exact CLI commands, and a failure-mode table keyed on real error codes. Fully executable and covering the common cases, matching anchor 5. | 5 / 5 |
Workflow Clarity | The Required Flow is a clearly numbered 0-12 sequence with explicit validation (step 0 environment probe with STOP conditions, "Verify: managePgDatabase(action=listMigrations)", post-DDL schema re-query with column comparison, "Verify both layers before claiming done"), feedback loops ("retry SQL at most twice" then query objects; MIGRATION_TASK_TIMEOUT handled via describeMigrationTask before any re-push), and a Quick Checks checklist — the anchor 5 pattern for database operations. | 5 / 5 |
Progressive Disclosure | All 10 referenced files exist in references/, are one level deep, clearly signaled with purpose (e.g. "MUST read before writing any bucket / upload / URL code"), and backed by a genuine routing index (references/index.md with a read order). However, substantial detail is inlined in SKILL.md — the storage failure-mode cheat sheet (duplicated by references/storage-pg.md), extended migration-action details, and the second copy of the API comparison table — so anchor 4's "minor organization gaps" fits better than anchor 5's "content appropriately split". | 4 / 5 |
Total | 17 / 20 Passed |