Svelte Llm MCP

What the svelte-llm-mcp MCP server does

The svelte-llm-mcp MCP server gives AI agents access to documentation for Svelte 5 and SvelteKit. Its documentation content is prepared through llmctx, a project that transforms technical documentation into condensed, LLM-friendly formats. The intended result is framework reference material that can be supplied to coding assistants while they work on Svelte applications.

This makes the server relevant when an agent needs framework-specific reference content and the project wants to use maintained documentation rather than an arbitrary collection of web pages.

How it works

llmctx organizes documentation through presets. A preset identifies a source repository, the files to include, and an optional prompt that can add context. File selection supports Markdown and MDX glob patterns, along with exclusions. Patterns are processed in order, allowing a preset to include broad documentation paths and then remove selected directories or files.

The svelte-llm-mcp MCP server exposes the resulting documentation context for use by AI coding assistants. The project also supports combining multiple documentation sources in one request; the README gives a Svelte and SvelteKit combination as an example. This is useful when a task spans the Svelte framework and its application framework rather than relying on one source alone.

Preset content is accessed through preset URLs, and the project points users to llmctx.com for the complete preset catalog.

Setup and configuration

It does describe how contributors add documentation presets: edit src/lib/presets.ts, add a preset entry with a title, repository owner, repository name, file globs, and optionally a prompt, then open a pull request.

That contribution workflow is relevant if the required framework documentation is not already represented by an available preset. New sources should be reliable and actively maintained.

Limitations and notes

It also does not name individual MCP tools, describe transport options, or state whether the server can modify source code. Treat it as a documentation-context service for AI assistants, not as a general Svelte project automation API.

Cursor and Zed are named as example AI coding assistants that can use the prepared documentation context.