菜单
Supabase PostgreSQL best practices library — 33 impact-rated rules (CRITICAL→LOW) covering connection pooling, RLS, locking, VACUUM, full-text search, and JSONB indexing. Use when writing, reviewing, or optimizing PostgreSQL queries, schema, or configuration.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/kumaran-is/claude-code-onboarding --skill postgres-best-practices命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | postgres-best-practices |
| description | Supabase PostgreSQL best practices library — 33 impact-rated rules (CRITICAL→LOW) covering connection pooling, RLS, locking, VACUUM, full-text search, and JSONB indexing. Use when writing, reviewing, or optimizing PostgreSQL queries, schema, or configuration. |
| argument-hint | [category or specific rule name] |
| allowed-tools | Read |
| context | fork |
| metadata | {"triggers":"connection pooling, pgbouncer, RLS row level security, SKIP LOCKED, queue workers, deadlock, advisory lock, VACUUM ANALYZE, full-text search tsvector, JSONB GIN index, pg_stat_statements, connection limits, max_connections, idle timeout, prepared statements, transaction mode pooling","related-skills":"database-schema-designer, sql-optimization-patterns, postgresql, sql-pro, vector-database","domain":"infrastructure","role":"optimizer","scope":"implementation","output-format":"document"} |
| last-reviewed | 2026-03-15 |
NEVER TUNE WHAT YOU HAVEN'T MEASURED — CHECK pg_stat_statements AND pg_stat_activity BEFORE CHANGING CONFIGURATION OR ADDING INDEXES
SKIP LOCKED and advisory lockstsvector/tsqueryjsonb_ops vs jsonb_path_ops)sql-optimization-patterns)database-schema-designer)sql-pro)| Priority | Category | Impact | Rules |
|---|---|---|---|
| 1 | Query Performance | CRITICAL | rules/query-*.md (5 rules) |
| 2 | Connection Management | CRITICAL | rules/conn-*.md (4 rules) |
| 3 | Security & RLS | CRITICAL | rules/security-*.md (3 rules) |
| 4 | Schema Design | HIGH | rules/schema-*.md (6 rules) |
| 5 | Concurrency & Locking | MEDIUM-HIGH | rules/lock-*.md (4 rules) |
| 6 | Data Access Patterns | MEDIUM | rules/data-*.md (4 rules) |
| 7 | Monitoring & Diagnostics | LOW-MEDIUM | rules/monitor-*.md (3 rules) |
| 8 | Advanced Features |
| LOW |
rules/advanced-*.md (2 rules) |
rules/conn-pooling.md — PgBouncer setup, pool size formula (cores × 2) + spindles, transaction vs session moderules/conn-limits.md — max_connections calculation, prevent database OOM crashrules/conn-idle-timeout.md — Reclaim 30-50% idle connection slotsrules/conn-prepared-statements.md — Prepared statement conflicts in transaction-mode poolingrules/lock-skip-locked.md — Atomic claim-and-update queue pattern, 10x worker throughputrules/lock-deadlock-prevention.md — Consistent lock ordering, deadlock prevention patternsrules/lock-short-transactions.md — Minimize transaction duration, 3-5x throughputrules/lock-advisory.md — Application-level coordination without row-level locksrules/security-rls-basics.md — ALTER TABLE ... ENABLE ROW LEVEL SECURITY + policy patternsrules/security-rls-performance.md — SECURITY DEFINER functions, 5-10x faster RLS queriesrules/security-privileges.md — Principle of least privilege, specific GRANT patternsrules/monitor-vacuum-analyze.md — VACUUM/ANALYZE patterns, autovacuum tuningrules/monitor-pg-stat-statements.md — Enable and query pg_stat_statementsrules/monitor-explain-analyze.md — EXPLAIN ANALYZE usage guiderules/advanced-full-text-search.md — tsvector, ts_rank, GIN full-text index, 100x vs LIKErules/advanced-jsonb-indexing.md — GIN index, jsonb_ops vs jsonb_path_ops trade-offrules/query-missing-indexes.md — EXPLAIN Seq Scan detectionrules/query-index-types.md — B-tree/GIN/BRIN/Hash selectionrules/query-composite-indexes.md — Leftmost prefix rulerules/query-covering-indexes.md — INCLUDE columns for Index Only Scanrules/query-partial-indexes.md — Filtered index subsetsrules/schema-primary-keys.md — IDENTITY vs UUID vs UUIDv7rules/schema-data-types.md — Forbidden types, correct alternativesrules/schema-foreign-key-indexes.md — Diagnostic SQL to find all missing FK indexesrules/schema-lowercase-identifiers.md — snake_case, no quoted identifiersrules/schema-partitioning.md — Range partitioning patternsrules/schema-foreign-key-indexes.md — FK index enforcementmin_size/max_size configuration (see rules/conn-pooling.md)app.current_user_id session variable before queriesDATABASE_URL=...?connection_limit=10&pool_timeout=20 (see rules/conn-pooling.md)pgbouncer=true parameterprisma.$queryRaw for queue worker queriesr2dbc.pool.max-size, r2dbc.pool.initial-size (see rules/conn-pooling.md)ConnectionFactory with per-request session variable injectiondatabase-schema-designer — Schema design, Flyway migrations, normalizationsql-optimization-patterns — EXPLAIN analysis, N+1, cursor pagination, batch opspostgresql — PostgreSQL-specific types, gotchas, extensions, JSONB depthsql-pro — Cloud analytics, HTAP, BigQuery/Snowflakevector-database — pgvector HNSW/IVFFlat alignment