hyperterse
The MCP framework. Connect your data to your agents.
claude mcp add --transport stdio hyperterse-hyperterse hyperterse start \ --env HTS_PORT="Port to run the HTTP MCP runtime (default 8080)" \ --env HTS_CONFIG="Path to your .hyperterse config or defaults"
How to use
Hyperterse is a tool-first MCP framework that turns tool configurations into callable tools powered by a declarative runtime. It discovers tools and adapters from a project scaffold, then exposes an MCP runtime over HTTP so clients can invoke tools via JSON-RPC like calls. You can model data flows with root configs, adapter files, and per-tool configurations, and extend execution with input transforms, authentication, and output transforms. The system supports script hooks and sandboxed embedded scripting to customize tool behavior. To get started, install the Hyperterse CLI, initialize a project, and run the server; you’ll be able to list, invoke, and transform tool results through HTTP endpoints such as /heartbeat and /mcp.
Once running, you can interact with tools by sending JSON-RPC requests to the MCP endpoint (for example, invoking tools/list). Tools can be backed by databases via adapters or be script-backed with custom handlers. This provides a consistent, declarative approach to exposing data-enabled tools while preserving the ability to inject transforms, auth layers, and adapters as needed.
How to install
Prerequisites:
- A system with curl or a web installer, and a supported shell (bash/zsh).
- Network access to fetch the Hyperterse CLI installer.
Step 1: Install Hyperterse
- Run the official installer script:
curl -fsSL https://hyperterse.com/install | bash
Step 2: Initialize a project
- Create a new Hyperterse project scaffold (or run inside an existing repo):
hyperterse init
This will scaffold the project structure including the .hyperterse root config, adapters, and tool templates.
Step 3: Configure tools and adapters
- Add or edit files under app/tools and app/adapters according to your data sources and tool definitions. Each tool typically has a config.terse and optional input/output hooks.
Step 4: Run the MCP runtime
- Start the server in development mode:
hyperterse start
To run with automatic rebuilds on changes, use:
hyperterse start --watch
Optional: validate and build for production
hyperterse validate
hyperterse build -o dist
hyperterse serve dist/
Additional notes
Tips and notes:
- The root configuration lives in .hyperterse; ensure adapters (.terse) and tool configs (.terse) reference the correct paths.
- Security: Hyperterse validates inputs by type, but SQL/statement placeholders are substituted via raw string replacement; enforce strict input constraints and safe query patterns in production.
- If you customize tool behavior with scripts, they run in a sandboxed runtime compiled into the binary; manage script hooks carefully and test with hyperterse validate.
- For local development, the default HTTP endpoint is http://localhost:8080; you can override ports with environment variables (HTS_PORT).
- Use the /heartbeat endpoint to check liveness and /mcp to invoke tool actions via JSON-RPC requests.
Related MCP Servers
pgmcp
An MCP server to query any Postgres database in natural language.
MCP-PostgreSQL-Ops
🔍Professional MCP server for PostgreSQL operations & monitoring: 30+ extension-independent tools for performance analysis, table bloat detection, autovacuum monitoring, schema introspection, and database management. Supports PostgreSQL 12-17.
mcp-dap
MCP server to communicate with DAP servers allowing AI Agents the ability to debug live programs.
mcp-konnect
A Model Context Protocol (MCP) server for interacting with Kong Konnect APIs, allowing AI assistants to query and analyze Kong Gateway configurations, traffic, and analytics.
symfony
A Symfony package designed for building secure servers based on the Model Context Protocol, utilizing Server-Sent Events (SSE) and/or StreamableHTTP for real-time communication. It offers a scalable tool system tailored for enterprise-grade applications.
mcp-framework
Rust MCP framework for building AI agents