alonso-skills/postgres
Comprehensive PostgreSQL reference for developers and DBAs covering versions 14–18. Use whenever the user asks about PostgreSQL syntax, DDL/DML/DQL, joins, LATERAL, CTEs, window functions, GROUPING SETS, DISTINCT ON, RETURNING, ON CONFLICT, PL/pgSQL, functions, procedures, triggers, views, materialized views, indexes (B-tree/GIN/GiST/BRIN/Hash/Bloom), MVCC, VACUUM, autovacuum, WAL, TOAST, partitioning, replication (streaming/logical), backup, PITR, HA (Patroni/repmgr), pgBouncer, EXPLAIN ANALYZE, RLS, roles, extensions (pgvector, PostGIS, TimescaleDB, Citus, pg_trgm, pg_cron), JSON/JSONB, full-text search, UUID, timestamptz, COPY, system catalogs, collations, large objects, cursors, GUC, or any Postgres administration, performance, security, replication, backup, or recovery topic.
94
94%
Does it follow best practices?
Impact
94%
1.36xAverage score across 3 eval scenarios
Passed
No known issues
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is an excellent skill description that thoroughly covers what the skill does (comprehensive PostgreSQL reference for versions 14-18) and when to use it (explicit 'Use whenever...' clause with an exhaustive list of trigger terms). The description is highly specific, uses natural keywords that users would actually say, and is clearly distinguishable from other database or coding skills. The only minor concern is that the sheer length could be slightly unwieldy, but the content is substantive rather than padded.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description lists an extensive set of concrete capabilities and topics: DDL/DML/DQL, joins, LATERAL, CTEs, window functions, specific index types, MVCC, VACUUM, partitioning, replication types, backup, PITR, and many more. This goes well beyond naming a domain and some actions. | 3 / 3 |
Completeness | Clearly answers both 'what' (comprehensive PostgreSQL reference covering versions 14-18) and 'when' with an explicit 'Use whenever the user asks about...' clause followed by an exhaustive list of trigger topics. | 3 / 3 |
Trigger Term Quality | Excellent coverage of natural terms users would say: 'PostgreSQL syntax', 'window functions', 'EXPLAIN ANALYZE', 'pgvector', 'PostGIS', 'JSON/JSONB', 'full-text search', 'autovacuum', 'replication', 'backup', 'Postgres'. The description includes both formal terms and common abbreviations users would naturally use. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive — it is clearly scoped to PostgreSQL specifically (not generic SQL or databases), mentions PostgreSQL-specific features (DISTINCT ON, MVCC, WAL, TOAST, PL/pgSQL, Patroni, pgBouncer), and version ranges. Very unlikely to conflict with skills for other databases or general coding. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-architected skill that serves as an effective routing layer for a massive PostgreSQL knowledge base. Its greatest strengths are the comprehensive routing table with clear keyword matching, the explicit multi-step workflow, and the excellent progressive disclosure pattern. Minor weaknesses include some redundancy between sections and the routing/disambiguation tables partially overlapping, though the disambiguation table adds genuine value with its 'why' column explaining cross-file relationships.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The routing table and disambiguation tips are well-structured and earn their tokens, but there's some redundancy (e.g., the disambiguation table partially restates routing table information). The versioning/provider neutrality section repeats guidance already stated in the intro. The response format section is moderately verbose but mostly justified given the complexity of the domain. | 2 / 3 |
Actionability | The workflow is concrete and executable: identify topic → load specific reference file via Read tool → follow See Also → answer with direct code/SQL first, then version notes, then citations. The routing table provides exact file paths and keyword matching. The response format specifies precise patterns (version admonitions, EXPLAIN reading instructions, managed-service phrasing). This is highly actionable guidance for Claude. | 3 / 3 |
Workflow Clarity | The 6-step usage workflow is clearly sequenced with explicit decision points (step 6: when to ask user about version). The disambiguation table provides clear fallback logic for ambiguous queries. The NOTE callout provides cross-cutting entry points for questions that don't match a single file. The instruction to never answer from memory alone when a reference exists is a strong validation checkpoint. | 3 / 3 |
Progressive Disclosure | This is an exemplary progressive disclosure structure: SKILL.md serves purely as a routing layer with a comprehensive keyword-to-file mapping table covering 102 reference files. All executable detail lives in the referenced files. Navigation is one level deep (SKILL.md → reference file, with See Also for cross-references). The disambiguation tips table elegantly handles multi-file routing. However, no bundle files were provided to verify the references actually exist, but the structure itself is well-designed. | 3 / 3 |
Total | 11 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
Reviewed
Table of Contents