I'm Samrith, the founder of Hyperterse.

We built Hyperterse because I kept running into the same problem when building AI agents: safely connecting databases to LLMs requires writing endless API endpoints, managing boilerplate validation, and building custom integrations for each AI framework, build custom MCP integrations for each AI framework, or worry about exposing SQL or connection strings to clients.

We built a runtime server that consolidates this into a single declarative configuration file. You define your queries once, and Hyperterse automatically generates typed REST endpoints, MCP tools, OpenAPI specs, and LLM-friendly documentation.

Instead of writing API endpoints, you define queries in a simple config file. Hyperterse handles:

- Query Definition: Write queries once with typed inputs

- Endpoint Generation: Each query becomes a REST endpoint

- MCP Integration: Queries are automatically exposed as MCP tools for AI agents

- Documentation: OpenAPI 3.0 specs and LLM-readable docs generated automatically

- Security: SQL and connection strings stay server-side—never exposed to clients

Here's what a typical query looks like. This replaces about 500-1000 lines of boilerplate API code:

adapters:

  my_db:

    connector: postgres

    connection_string: "postgresql://user:pass@localhost:5432/db"

  get-user:

    use: my_db

    description: "Retrieve a user by email"

    statement: |

      SELECT id, name, email, created_at

      FROM users

      WHERE email = {{ inputs.email }}

    inputs:

      email:

        type: string

        description: "User email address"

You can use Hyperterse for a variety of use cases, with the most prominent being:

- AI agents and LLM tool calling

- RAG applications with structured database queries

- Rapid API prototyping

- Multi-agent systems sharing database access