Model Context Protocol

The Model Context Protocol (MCP) is an open standard for connecting AI agents to external systems. With MCP, Syntic Code can reach beyond your filesystem and shell to query databases, read from ticketing systems, call internal APIs, and pull in live documentation, all through a uniform interface. An MCP server exposes a set of capabilities; Syntic Code acts as the client that discovers and uses them, giving Amara real context about the systems your code touches.

What MCP gives you

An MCP server can offer three kinds of capabilities. Tools are actions the agent can invoke, such as running a query or creating a ticket. Resources are readable data the agent can pull into context, like a document or a record. Prompts are reusable templates a server can supply to guide the agent. Syntic Code negotiates these capabilities when it connects, so a single configuration entry can unlock a whole toolbox for Amara.

How Syntic Code uses MCP

You declare servers in your Syntic Code configuration, giving each one a name, a transport, and any credentials it needs. On startup, syntic connects, lists what each server offers, and makes those tools and resources available to the agent during your session. Because servers are configured per project or per user, teams can share a common set of connections by committing them to the repository, while individuals keep private servers in their user config.

Where to go next

Read the Quickstart to connect your first MCP server end to end, from installing a server to invoking one of its tools in a session. Then use the Reference for the full configuration schema and the details of each transport, stdio, SSE, and HTTP, so you can wire up local processes and remote services alike.