sublime_mcp_server
An MCP server that exposes Sublime Text's unsaved buffers as tools via local HTTP, enabling AI clients to read, list, and close buffers without saving to disk.
README
MCPServer
MCPServer is a Sublime Text package that exposes MCP tools over local HTTP from inside Sublime's Python runtime.
Today it ships with MCP tools for unsaved buffers:
list_open_buffersread_bufferclose_buffer
Use Cases
- Read temporary scratch notes or prompts directly from open unsaved tabs.
- Summarize or refine a draft in an AI client, then send the result to another MCP such as Notion.
- Clean up working tabs by closing consumed scratch buffers via
close_buffer. - Build lightweight editor automations without saving throwaway notes to disk.
The architecture is fully in-process. By default it starts:
- Sublime bridge on
127.0.0.1:6123 - MCP endpoint on
127.0.0.1:6124/mcp
No separate Node or TypeScript MCP server is required.
Requirements
- Sublime Text 4 build
4171or newer - Python 3.8 compatibility for the plugin package
- Python 3.11+ if you want to run the repo tests locally
- A local MCP client that can talk to a streamable HTTP MCP endpoint
Project Layout
sublime_mcp_server/: bridge logic and in-process MCP HTTP serversublime_mcp_server_plugin.py: Sublime Text plugin entrypointmessages.jsonandmessages/*.txt: Package Control install and upgrade messagespython_tests/: pytest coverage for bridge and MCP behavior
Install The Sublime Plugin
For normal usage, install the packaged artifact into Sublime's Installed Packages folder.
On macOS:
mkdir -p "$HOME/Library/Application Support/Sublime Text/Installed Packages"
cp dist/MCPServer.sublime-package \
"$HOME/Library/Application Support/Sublime Text/Installed Packages/MCPServer.sublime-package"
If you previously installed an unpacked development copy, remove it first so Sublime does not load the package twice:
rm -rf "$HOME/Library/Application Support/Sublime Text/Packages/MCPServer"
Restart Sublime Text after copying the files.
The packaged release includes the repository's tracked .python-version file at the package root so Sublime can select the correct Python runtime when loading the archive.
For local development, you can still use an unpacked checkout under Packages/MCPServer, but that is now a development-only workflow instead of the recommended installation method.
Port Configuration
The package reads its host and port configuration from MCPServer.sublime-settings.
Default values:
{
"bridge_host": "127.0.0.1",
"bridge_port": 6123,
"mcp_host": "127.0.0.1",
"mcp_port": 6124
}
To override them in Sublime Text:
- Open the command palette.
- Run
Preferences: MCPServer Settings. - Put your overrides in the right-hand user settings pane.
- Restart Sublime Text.
Example:
{
"bridge_port": 7001,
"mcp_port": 7002
}
The bridge and MCP ports must be different when using the same host.
Package And Release
There are two common release paths for Sublime Text packages:
- Package Control default channel: publish this repo on GitHub, create a semantic version tag such as
0.2.5, and submit the repository to Package Control once. After approval, future semver tags are how users receive updates. - Custom distribution: build a
.sublime-packagearchive and host your ownpackages.jsonchannel if you do not want to use Package Control's default channel.
This repo now includes release-facing metadata:
- central package display name, MCP server name, version, and description in sublime_mcp_server/init.py
- Package Control install and upgrade messages in messages.json
- release notes for
0.2.5in messages/0.2.5.txt
Suggested release checklist:
- Update the version in pyproject.toml and sublime_mcp_server/init.py.
- Add a matching upgrade note under
messages/. - Build a
.sublime-packageartifact from the repo root:
python3 scripts/build_release.py
- Run
pytestandpython3 -m py_compile sublime_mcp_server_plugin.py sublime_mcp_server/*.py. - Commit the release changes.
- Create and push a semver tag such as
0.2.5. - If this is the first public release, submit the GitHub repository URL to Package Control.
For manual installs, use the generated dist/MCPServer.sublime-package file. The stable filename keeps the package name clean in Sublime, and the builder injects package-metadata.json so the description and version are visible in package UIs.
Verify The Plugin Is Running
Once Sublime restarts, the plugin should start two localhost services:
GET http://127.0.0.1:6123/buffersPOST http://127.0.0.1:6124/mcp
Check the bridge:
curl -sf http://127.0.0.1:6123/buffers | python3 -m json.tool
Check the MCP endpoint:
curl -sf \
-X POST http://127.0.0.1:6124/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"1.0.0"}}}' \
| python3 -m json.tool
You should see serverInfo.name set to sublime_mcp_server.
If you changed the ports in settings, use those values instead of 6123 and 6124.
MCP Client Setup
This project serves MCP over HTTP from inside Sublime, so your MCP client only needs to connect to the plugin's endpoint.
Recommended MCP client configuration:
{
"mcpServers": {
"sublime_mcp_server": {
"type": "http",
"url": "http://127.0.0.1:6124/mcp"
}
}
}
Available Tools
list_open_buffers
Lists unsaved open Sublime buffers. Each buffer descriptor includes:
buffer_idtitlewindow_idis_dirtysyntaxpreview
read_buffer
Reads the current live text and metadata for one unsaved Sublime buffer.
Input:
{
"buffer_id": "38"
}
close_buffer
Force-closes an unsaved Sublime buffer and discards its current contents.
Input:
{
"buffer_id": "38"
}
Manual End-To-End Check
- Open Sublime Text.
- Create one or more new unsaved tabs.
- Add some test text.
- Verify the buffer list:
curl -sf http://127.0.0.1:6123/buffers | python3 -m json.tool
- Use an MCP client or direct JSON-RPC request to list tools and read a buffer.
Direct JSON-RPC example:
curl -sf \
-X POST http://127.0.0.1:6124/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' \
| python3 -m json.tool
curl -sf \
-X POST http://127.0.0.1:6124/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"read_buffer","arguments":{"buffer_id":"38"}}}' \
| python3 -m json.tool
curl -sf \
-X POST http://127.0.0.1:6124/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":4,"method":"tools/call","params":{"name":"close_buffer","arguments":{"buffer_id":"38"}}}' \
| python3 -m json.tool
Development
Run tests:
pytest
Check syntax:
python3 -m py_compile sublime_mcp_server_plugin.py sublime_mcp_server/*.py
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.