pgContext User Guide
pgContext is an open-source PostgreSQL extension for AI vector and hybrid retrieval.
This guide distinguishes stable, implemented behavior from experimental and planned paths. pgContext 0.3.0 targets PostgreSQL 17 and 18.
Current Status
The Supported Features page is the canonical inventory of currently implemented behavior and its maturity. It is checked against the code-backed capability contract. Planned-only work remains in the post-V1 product roadmap.
Implemented Core Behavior
- Supported features
- Installation
- Configuration
- Packaged playground
- Collections
- Production quickstart
- SQL API contract
- Dense vectors and exact search
- Multi-tenancy runbook
- Client-facing examples
- Filters
- Hybrid retrieval
- Lexical and fuzzy retrieval
- Adaptive-dimension retrieval
- Multi-model retrieval
- Semantic reranking
- Automatic document chunking
- Exact-first readiness
- Retrieval methods overview
- Indexes
- Rebuildable storage artifacts
- Operations and support
- Troubleshooting and maintenance runbook
- Known limitations
- pgvector and Qdrant parity matrix
- Exact metric and operator matrix
- Metric definitions and edge-case semantics
- Post-V1 product roadmap
- Installed SQL object and option inventory
- Migrating from pgvector
- Support, version, upgrade, and deprecation policy
- Rollback and repair plan
- First production release notes
- Error categories and SQLSTATEs
- Security model
PostgreSQL Support
PostgreSQL 17 and 18 are supported V1 release targets. OCI images are built and runtime-verified for linux/amd64 and linux/arm64; PostgreSQL 17 remains the primary benchmark and deep-lifecycle qualification target.