Releases: indiekitai/pg-dash
Releases · indiekitai/pg-dash
Release list
v0.13.0 — pgvector Health Checks + Streamable HTTP
What's New
🧬 pgvector Health Checks
New pgvector CLI subcommand and pg_dash_pgvector MCP tool:
pg-dash pgvector postgres://...- Checks pgvector installation and version
- Lists vector columns with dimension analysis (flags >2000d)
- Analyzes vector index types (IVFFlat vs HNSW) with parameter tuning
- IVFFlat
listsvalidation against row count (should be ~√rows) - HNSW
ef_constructionvsmratio check - Detects missing vector indexes on large tables
- Index size vs table size ratio warnings
🌐 MCP Streamable HTTP Transport
pg-dash MCP server now supports remote deployment as a shared team server:
# Run as HTTP server (stateless, MCP 2026-07-28 compatible)
pg-dash-mcp postgres://... --http
# Custom port
MCP_PORT=9000 pg-dash-mcp postgres://... --httpUses Hono + Web Standard StreamableHTTPServerTransport, compatible with the upcoming MCP 2026-07-28 stateless protocol spec.
Other Changes
- MCP tools count: 25 → 26
- Updated README with SQL safety section and pgvector documentation
v0.12.0 — Latest Stable
v0.12.0
Latest stable release. Includes all features from v0.8.0–v0.11.0:
- 25 MCP tools for AI agent integration
- Slow query analysis from pg_stat_statements
- Table & index bloat detection with autovacuum health
- Migration safety checker — catches lock risks before production
- Multi-environment schema diff — compare local vs staging vs prod
- CI integration —
--ciflag for GitHub Actions annotations - Prepared-statement cache warnings
npx @indiekitai/pg-dash check postgres://user:pass@host/dbv0.11.0 — Consolidated Tool
What's Changed
- pg-dash is again the single consolidated tool — Reverted the experimental pg-health split. All CLI subcommands, MCP server, and web dashboard remain in this one package.
- One install, one upgrade path:
npm install -g @indiekitai/pg-dash
v0.10.0 — Prepared Statement Cache Warnings
What's New
- Prepared-statement cache invalidation warning — Detects when prepared statement cache invalidation rate is high, which can cause performance issues in connection poolers (PgBouncer, Supavisor)
- Improved pg_stat_statements data handling
v0.9.1 — Bug Fix
Bug Fix
- Fixed unused indexes query that could miss certain index types
v0.9.0 — Table & Index Bloat Detection
What's New
- New
bloatcommand — Detect table and index bloat (dead tuples) that slow down queries and waste disk space.
npx @indiekitai/pg-dash bloat postgres://...- Shows dead tuple ratio per table, flags tables with ≥10% bloat
- Surfaces
last_autovacuumandlast_vacuumtimestamps - MCP tool:
pg_dash_bloat
v0.8.0 — Slow Query Analysis
What's New
- New
slow-queriescommand — Analyze slow queries frompg_stat_statements. Surface the most expensive queries by total time, mean time, or call count.
npx @indiekitai/pg-dash slow-queries postgres://... --limit 20 --min-calls 5- MCP tool:
pg_dash_slow_queries - Supports
--format jsonfor CI integration
v0.7.1 - Batch Operations & Improved Advisor
Whats New
Batch Operations
- pg_dash_fix now supports multiple statements: VACUUM t1; VACUUM t2; ANALYZE t3
- ALTER TABLE SET support for storage parameters
Improved Advisor
- pg_stat_statements detection - warns when not installed
- Small table autovacuum check (<50 rows)
- Fixed dead tuple percentage calculation
- Unused index threshold reduced from 1MB to 8KB
Bug Fixes
- pg_dash_explain parameter name fixed (query -> sql)
Full changelog: https://github.com/indiekitai/pg-dash/blob/master/CHANGELOG.md
v0.7.0 - AI-Powered Database Context
Whats New
AI-Powered Database Context
- fetch_db_context MCP tool - Get comprehensive database context
- pg_dash_query_natural MCP tool - Query database using natural language
- pg-dash check --ai-suggest: AI-powered fix suggestions
- pg-dash diff-env --ai-explain: LLM explains schema differences
详见 CHANGELOG.md