Conciseness

The body is dense with platform-specific facts (LFC/prefetch EXPLAIN options, pooled-vs-direct failure signatures, diagnostic distinctions between `long-running-queries`, `outliers`, and `calls`) rather than generic Postgres tutorials, so it is above the anchor-3 "some unnecessary explanation" level. It is not 5 because of minor padding: the lakebase-architecture intro paragraph, the Databricks aside, and "It's fully compatible with Postgres and works with any language, framework, or ORM" restate things Claude can infer or that the parent skill covers.

Actionability

Guidance is copy-paste executable: exact commands (`neon skills -s neon -y`, `neon inspect db <check>`, `neon env pull`), a complete SQL statement (`EXPLAIN (ANALYZE, BUFFERS, PREFETCH, FILECACHE)`), env-var names (`DATABASE_URL` vs `DATABASE_URL_UNPOOLED`), driver-specific notes with install commands, and concrete error signatures users will actually hit (`prepared statement "s0" already exists`, `SQLSTATE 25006`). This matches anchor 5: specific, ready-to-run examples covering the common cases.

Workflow Clarity

Both the Setup Flow (numbered 1-4) and the Performance Workflow (numbered 1-5) are clearly sequenced, and the database-specific risk cap is satisfied: step 5 says "Re-run the same check and workload to verify the change" (explicit feedback loop), "Interpret Results Safely" adds ask-before-mutating checkpoints ("ask before running the suggested `CREATE EXTENSION` statement", confirm before `VACUUM FULL`), and error recovery is spelled out ("retry the relevant check with an explicit `databaseName` to isolate it"). Anchor 5 fits; 4 would understate the validation coverage.

Progressive Disclosure

The Lakebase Search section cleanly offloads detail to four one-level-deep references that all exist in the bundle (references/vector-search.md, full-text-search.md, hybrid-search.md, lakebase-search-drizzle.md), and sibling skills (neon, neon-postgres-branches, postgres-best-practices) are signposted with fetch/install instructions. It is not 5 because the ~130-line troubleshooting/diagnostics section — candidate material for its own reference file, like the search material got — is fully inlined in SKILL.md.

Specificity

The description enumerates a comprehensive, specific scope ("connections, pooled vs direct, schema migrations, branching, autoscaling, scale-to-zero, instant restore, read replicas, IP allow lists, logical replication, and Lakebase Search") but frames it as "Guides and best practices for working with", which lists topic domains rather than the concrete actions of the anchor-5 example ("Extract text and tables from PDF files, fill forms, merge documents"). It sits above anchor 4 only in coverage, not in action-verb concreteness, so 4 fits best.

Completeness

Both what and when are explicitly answered: the what is the enumerated Lakebase Postgres capability list, and the when is the explicit clause "Use when the work is an existing DATABASE_URL, SQL, schema, inspect, or search" plus "Also use for ..." with concrete trigger phrases — the exact structure of the anchor-5 example. It is not 4 because the when-clause is fully explicit rather than merely present.

Trigger Term Quality

Natural trigger terms are comprehensive and include synonyms, exact package names, and a CLI subcommand users would literally type: "DATABASE_URL", "SQL", "schema", "inspect", "@neondatabase/serverless", "@neondatabase/neon-js", "neon inspect db", "semantic search", "vector search", "full-text search", "BM25", "hybrid search". This matches the anchor-5 example's coverage of natural terms, synonyms, and extensions; no common variation is missing.

Distinctiveness Conflict Risk

It carves out a clear niche (Lakebase Postgres on Neon) and actively reduces conflict risk by routing adjacent work elsewhere: "New backends, Auth, files, Functions, and LLM calls go to the parent `neon` skill." Voice is third person ("Guides", "go to"), so the voice penalty does not apply. Minimal overlap risk remains.