Bitbucket MCP Server
An MCP server for Bitbucket that provides pull request context to LLMs for automated code reviews.
🌬️ bitbucket-mcp-server
Simple MCP implementation for (mostly) pull requests on Bitbucket.
Provide context from your PRs to your favorite LLM and let it review them for you!
Written in Bun and Civet using Cursor as an assistant.
Read the documentation here.
🚧 This repository is a MCP learning exercise and it is still work in progress.
The API here lacks many features as of now. More features might be added in the near future.
Also, the documentation was written by Cursor itself. Don't expect it to be perfect. 🚧
Install
First clone this repository somewhere you'd like.
Then there are 2 ways to run it locally:
- If you want to download and use the release version, you can run via the binary available in this repository Releases;
- Otherwise, you'll have to either build from scratch, or install Bun and run the
index.civetfile directly.
Release
To run from the pre-built binaries, go to the Releases page and get the latest release from there.
Extract the binary somewhere you'd like.
Then, add this to your MCP servers settings:
{
"mcpServers": {
"Bitbucket": {
// Or just bitbucket-mcp-server if you made it
// available as a system/user-wide binary
"command": "/path/to/bitbucket-mcp-server",
"args": ["username:password"]
}
}
}
This is enough to get the server going.
Bun/Build
If you want to run with Bun (or build with it), you'll first need to install bun by running:
curl -fsSL https://bun.sh/install | bash
Then from the project directory install all dependencies with:
bun i
Running with Bun
To run with bun directly, add this entry to your MCP servers settings:
{ "mcpServers": { "Bitbucket": { "command": "bun", "args": [ "run", "/path/to/repository/src/index.civet", "username:password" ] } } }
Building
Run the build command from the project root directory:
bun run build
This will output a binary in the dist/ folder to whatever architecture you're using.
Then follow the same steps from the Release option to add the MCP server to your settings.
Development
Make sure you have both Bun's runtime and Civet's VSCode/Cursor extension installed.
Then after cloning the repository, install its dependencies:
cd /path/to/repository bun i
The best way to visualize everything working is by running:
bun inspect username:password
This will open the MCP Inspector with a GUI to play around and explore the MCP tools.
MCP tools limitation
MCP has a limitation of 40 tools that will be sent to the agent.
Currently bitbucket-mcp-server provides 24 tools total. Consider disabling the ones you don't use often to save some slots.
To do
- Maybe implement OAuth
This project was created using bun init in bun v1.2.17. Bun is a fast all-in-one JavaScript runtime.
相關伺服器
GitHub MCP Server
Integrate GitHub functionalities into AI assistants using the GitHub CLI.
Git MCP Server
An MCP server for performing Git operations and integrating with GitHub.
Feather Code MCP
A portable MCP server for performing various GitHub operations on any repository.
Git
Tools to read, search, and manipulate Git repositories
Obsidian GitHub MCP
Connects AI assistants to Obsidian vaults stored in GitHub repositories, enabling them to read, search, and analyze your notes and documentation.
GitLab CLI MCP Server
Seamlessly integrate AI assistants with GitLab using the glab CLI tool.
MCP Git Repo Browser
A server for browsing and interacting with local Git repositories.
Repomix
Packs repository contents into a single file for AI consumption.
GitHub
Interact with GitHub repositories, including file operations, repository management, and search.
Feather Code
A lightweight MCP server for interacting with GitHub repositories, requiring authentication.