pgEdge MCP Server for Postgres Now GA, Advancing Enterprise AI Infrastructure
Full featured Postgres MCP Server supports new and existing databases, to scale from prototype to production
pgEdge, the leading open-source enterprise Postgres company, today announced general availability (GA) of the pgEdge MCP Server for Postgres. This full featured and production ready MCP server is designed for developers building Agentic AI applications in environments with strict requirements for high availability, security, data sovereignty and/or global deployment.
Unlike other available MCP servers, the pgEdge MCP Server for Postgres works with new and existing databases running any standard version of Postgres (v14 and newer), and offers flexible deployment options including on-premises (even air-gapped), self-managed cloud, or a managed cloud service via pgEdge Cloud. This deployment flexibility, coupled with the enterprise-grade capabilities of pgEdge Enterprise Postgres, allows developers to take their agentic AI applications all the way from prototyping to running in production on compliant and secure infrastructure.
Read More: Domain-Specific Language Models: Is It the Next Evolution of Enterprise AI?
“Most MCP servers today aren’t built with enterprise requirements in mind,” said David Mitchell, CEO of pgEdge. “We designed the pgEdge MCP Server for Postgres to deliver the flexibility of open source with the performance, security, and deployment control enterprises need—whether that’s in the cloud, on-prem, or in air-gapped environments. It’s a critical step toward making AI infrastructure production ready.”
The pgEdge MCP Server for Postgres works with AI application builders and code generators such as Claude Code, Cursor, Windsurf, and VS Code Copilot, and supports frontier models from OpenAI and Anthropic in addition to locally hosted models using Ollama, LM Studio, and other OpenAI API compatible products.
As with other pgEdge Postgres products, the pgEdge MCP Server for Postgres is fully open source via the Postgres license,and is fully supported by pgEdge’s leading team of Postgres contributors and expert developers.
Key features include:
- Full schema introspection: The MCP server pulls detailed information about database structure beyond just table and column names, including primary keys, foreign keys, indexes, column types, and constraints. This allows the LLM to reason about the data model versus blindly querying it.
- Read-only and read/write support: Configurable via a switch.
- Security built-in: Supports stdio, HTTP and HTTPS, with TLS support, user and token authentication, and read-only enforcement by default.
- Multi-database support: Supports multiple databases (e.g. dev, staging, and production) and the ability to switch between them.
- Performance metrics: The server exposes pg_stat_statements and other data so the LLM can make performance optimization recommendations.
Read More: ITTech Pulse Exclusive Interview with Debbie Gordon Founder/CEO at Cloud Range
Additionally, this GA release adds the following enhancements:
- Custom tools which can be written in SQL, Python, Perl or Javascript.
- A DBA Toolkit with pre-defined tools for analyzing database health, identifying top resource-consuming queries, and making index recommendations.
- Configurable base URLs for LLM and embedding providers for deployments where access is managed via proxy servers.
- Multi-host connection support for HA and failover.
- Reduced token usage via use of tab separated values (TSV) versus JSON, pagination of results and context window compaction.
- Token management tools for better rate limit handling
Write to us [wasim.a@demandmediaagency.com] to learn more about our exclusive editorial packages and programmes.
