pgEdge Starfleet Adds MCP, RAG and Database Branching to Managed Postgres

By AiCybr Editorial · Published September 28, 2026

pgEdge launched Starfleet on September 28, 2026, as a Postgres platform for AI and enterprise applications with built-in Model Context Protocol (MCP) access, a RAG server and copy-on-write database branching. The managed service starts at $25 per database per month after a 14-day free trial, while the same pgEdge Postgres platform can also run in a customer's cloud or on self-managed infrastructure.

Starfleet Managed currently launches in AWS US East-2. The BYO Cloud option can run in AWS, Google Cloud or Microsoft Azure regions, and pgEdge also supports self-hosted and air-gapped deployments through its enterprise platform. pgEdge says the database layer remains standard community PostgreSQL, avoiding an application migration to a separate proprietary database when a prototype moves into a controlled production environment.

The practical focus is agent access to operational data. Each managed database includes an authenticated MCP endpoint and a RAG server, while database branches give parallel development or agent experiments isolated writable Postgres instances. pgEdge documents those branches as copy-on-write clones built on standard Postgres storage.

Starfleet deployment and pricing

Pricing is flat per database for the managed service and billed hourly while a branch exists, prorated from the selected instance size's monthly price. pgEdge's current pricing page lists the $25 monthly starting point and larger dedicated sizes through 64 vCPU and 256 GB of memory.

MCP access is built into each managed database

Starfleet exposes an MCP server so tools such as Claude Code, Cursor and Replit can work with database schema and data through an authenticated endpoint. pgEdge's broader MCP implementation supports PostgreSQL 14 and newer, including existing standard Postgres deployments and services such as Amazon RDS.

For read-oriented agent workflows, pgEdge pairs MCP access with SafeSession, which is designed to enforce read-only database sessions at the PostgreSQL layer. That is relevant when an agent needs schema inspection, query generation or analysis without receiving general write authority.

The RAG service adds retrieval over Postgres data using pgvector, hybrid search and related pgEdge tooling. This keeps transactional data, vector retrieval and agent-facing database access within the Postgres environment instead of requiring a separate vector database for the basic architecture.

Branches isolate agent experiments

Database branching is one of Starfleet's more useful development features. A branch creates an isolated writable Postgres instance using copy-on-write storage. Teams can give separate coding agents or test workflows their own database state, allowing schema changes, migrations and generated SQL to be evaluated away from the production database.

This model also addresses a common problem with parallel agent development: source-code branches are easy to isolate, while database mutations can collide when multiple agents share one development database. Matching an agent or feature branch with an isolated database branch gives each experiment its own mutable state.

Production deployment options

Starfleet Managed is the fastest path for a hosted database, while BYO Cloud moves the infrastructure into the customer's cloud account. pgEdge positions the latter for organizations that need tighter control over networking, data location and production infrastructure. Self-hosted deployments extend the same platform to on-premises and air-gapped environments.

For high-availability deployments, pgEdge offers automated failover and optional active-active multi-region replication through its Spock replication technology. The company's product material also describes zero-downtime maintenance for BYO Cloud and on-premises configurations.

The deployment model is the main architectural distinction: an application can begin on a small managed Postgres database with agent tooling, then move to customer-controlled infrastructure while retaining PostgreSQL semantics and the same pgEdge platform components.

What to evaluate before adopting Starfleet

Teams evaluating Starfleet should separate the database portability claim from the surrounding managed platform. Standard PostgreSQL compatibility reduces application-level migration work, while operational features such as branching, MCP, RAG, SafeSession, multi-region management and the managed control plane are pgEdge capabilities that should be evaluated against the team's security and infrastructure requirements.

The launch is particularly relevant for teams already standardizing on PostgreSQL and adding coding agents or RAG workflows. Its strongest proposition is consolidation: agent database access, retrieval, isolated experimentation and multiple production deployment models are available around the same Postgres base.

Sources

• pgEdge, Starfleet product page and current feature/deployment details: https://www.pgedge.com/products/starfleet
• pgEdge, current Starfleet pricing and deployment matrix: https://www.pgedge.com/pricing
• pgEdge documentation, Starfleet Managed and BYO Cloud overview: https://docs.pgedge.com/pgedge-starfleet/
• pgEdge, Postgres MCP Server general availability and PostgreSQL 14+ compatibility: https://www.pgedge.com/press-releases/pgedge-mcp-server-for-postgres-now-generally-available-including-in-pgedge-cloud-advancing-enterprise-ai-infrastructure
• CIO, independent September 28 coverage of the Starfleet launch and production positioning: https://www.cio.com/article/4226629/pgedge-launches-starfleet-to-help-developers-take-postgres-ai-apps-from-prototype-to-production-2.html