Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion documentation/docs/10-introduction/10-overview.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ The Svelte MCP ([Model Context Protocol](https://modelcontextprotocol.io/docs/ge
The setup varies based on the version of the MCP you prefer — remote or local — and your chosen MCP client (e.g. Claude Code, Codex CLI or GitHub Copilot):

- [local setup](local-setup) using `@sveltejs/mcp`
- [remote setup](remote-setup) using [mcp.svelte.dev/mcp](https://mcp.svelte.dev/mcp)
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't think this should be a link, even if it redirects to the MCP docs

- [remote setup](remote-setup) using `https://mcp.svelte.dev/mcp`

## Usage

Expand Down
2 changes: 1 addition & 1 deletion documentation/docs/20-setup/30-remote-setup.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
title: Remote setup
---

The remote version of the MCP server is available on `https://mcp.svelte.dev/mcp`.
The remote version of the MCP server is available at `https://mcp.svelte.dev/mcp`.

Here's how to set it up in some common MCP clients:

Expand Down
4 changes: 2 additions & 2 deletions documentation/docs/30-capabilities/10-tools.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
title: Tools
---

The following tools are provided by the MCP server to the model, which can decide to call one or more of them during a session:
The following tools are provided by the MCP server to the model you are using, which can decide to call one or more of them during a session:

## list-sections

Expand All @@ -14,7 +14,7 @@ Allows the model to get the full (and up-to-date) documentation for the requeste

## svelte-autofixer

Uses static analysis to provide suggestions for the generated code. It should be invoked in a loop by the model until all issues and suggestions are resolved.
Uses static analysis to provide suggestions for code that your LLM generates. It can be invoked in an agentic loop by your model until all issues and suggestions are resolved.

## playground-link

Expand Down
19 changes: 1 addition & 18 deletions documentation/docs/30-capabilities/30-prompts.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,21 +6,4 @@ This is the list of available prompts provided by the MCP server. Prompts are se

## svelte-task

This prompt should be used whenever you are asking the model to work on some Svelte-related task. It will instruct the LLM on which documentation sections are available, which tool to invoke, when to invoke it, and how to interpret the result. It will ask you for the description of the task and the returned value will look like this:
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't like putting the prompt verbatim in the docs, that's just asking to forget to update it. 😅 If we really want to we could link to the GitHub source code for the prompt.


```
You are a Svelte expert tasked to build components and utilities for Svelte developers. If you need documentation for anything related to Svelte you can invoke the tool \`get-documentation\` with one of the following paths:
<available-docs-paths>
[all available docs]
</available-docs-paths>

Every time you write a Svelte component or a Svelte module you MUST invoke the \`svelte-autofixer\` tool providing the code. The tool will return a list of issues or suggestions. If there are any issues or suggestions you MUST fix them and call the tool again with the updated code. You MUST keep doing this until the tool returns no issues or suggestions. Only then you can return the code to the user.

This is the task you will work on:

<task>
[your task here]
</task>

If you are not writing the code into a file, once you have the final version of the code ask the user if they want to generate a playground link to quickly check the code in it and if they answer yes call the \`playground-link\` tool and return the url to the user nicely formatted. The playground link MUST be generated only once you have the final version of the code and you are ready to share it, it MUST include an entry point file called \`App.svelte\` where the main component should live. If you have multiple files to include in the playground link you can include them all at the root.
```
This prompt should be used whenever you are asking the model to work on a Svelte-related task. It will instruct the LLM which documentation sections are available, which tools to invoke, when to invoke them, and how to interpret the results.