← Back to directory

Bitbucket Server MCP

Community

MCP server for managing Bitbucket Server pull requests

GitHub source repository ↗

★ 66 Stars Category · Dev Tools Popular Source revision f57cb8624786

56FMRS · C

Reliability

7/20

Security and permissions

9/20

Maintenance

11/20

Documentation

16/20

Setup experience

13/20

This server offers a fairly comprehensive toolset for the Bitbucket Server pull-request workflow — project/repository discovery, PR creation through merge, comments and batch reviews, code search, and file browsing — with a read-only mode and custom headers for enterprise network setups; credentials must be protected and the irreversibility of write actions outside read-only mode should be kept in mind.

Read the FMRS scoring method →

Bitbucket Server MCP is a Node.js MCP server that exposes the Bitbucket Server REST API for project and repository discovery, full pull-request lifecycle management (create, update, approve, merge, decline), code search, file reading, branch and commit history browsing, comment management, and CI/CD code insights (e.g. SonarQube). It supports configuring a default project, a read-only mode, and custom HTTP headers (useful for Zero Trust tokens or proxies) via environment variables.

Tools

list_projects

Lists all accessible Bitbucket projects with their details, useful for project discovery and finding correct project keys.

list_repositories

Browses repositories within a specific project or across all accessible projects, returning clone URLs and metadata.

create_pull_request

Creates a new pull request with source/target branches and reviewers, supporting cross-repository PRs from forks.

update_pull_request

Safely updates a pull request's title, description, or reviewers using a read-modify-write pattern that preserves unspecified fields.

get_pull_request

Retrieves detailed pull request information including status, reviewers, commits, and metadata.

merge_pull_request

Merges an approved pull request using merge-commit, squash, or fast-forward strategy.

decline_pull_request

Declines a pull request that should not be merged, optionally with a reason message.

add_comment

Adds a comment to a pull request, supporting Markdown, threaded replies, and open or pending (draft) state.

add_comment_inline

Adds an inline comment on a specific line of a pull request's diff.

get_diff

Retrieves the code diff for a pull request, with configurable context lines and per-file line limits for large diffs.

get_reviews

Fetches review history, approval status, and reviewer feedback for a pull request.

get_activities

Gets the complete activity timeline for a pull request, including comments, reviews, and commits.

get_comments

Filters pull request activities to return only comments, focusing on discussion content.

Searches code and filenames across repositories with project/repository filtering; only works on default branches.

get_file_content

Reads the content of a specific file from a repository, with pagination for large files.

browse_repository

Browses files and directories in a repository to explore its structure.

list_pull_requests

Lists pull requests in a repository, filterable by state, author, and direction.

list_branches

Lists repository branches, identifying the default branch and latest commit info per branch.

list_commits

Lists commit history for a repository, filterable by branch and author.

delete_branch

Deletes a branch from a repository, with a safety check preventing deletion of the default branch.

approve_pull_request

Approves a pull request as the current authenticated user.

unapprove_pull_request

Removes the current user's approval from a pull request.

edit_comment

Edits an existing comment's text, requiring the comment version for optimistic locking.

delete_comment

Deletes a comment from a pull request, requiring the comment version for optimistic locking.

publish_review

Publishes all pending draft comments at once, optionally setting an approval status and overview comment.

get_code_insights

Fetches Code Insights reports (e.g. SonarQube, security scans) and their annotations for a pull request.

get_dashboard_pull_requests

Lists pull requests across all repositories for the authenticated user, filterable by role and status.

Setup

Ensure Node.js >= 16 is installed.
Optionally install via Smithery (npx -y @smithery/cli install @garc33/bitbucket-server-mcp-server --client claude), or install manually with npm install followed by npm run build.
Add a server entry to your MCP client configuration, setting command to node and args to the path of the built build/index.js.
Set required environment variables: BITBUCKET_URL, plus either BITBUCKET_TOKEN or BITBUCKET_USERNAME/BITBUCKET_PASSWORD for authentication; optionally set BITBUCKET_DEFAULT_PROJECT, BITBUCKET_READ_ONLY, BITBUCKET_CUSTOM_HEADERS, BITBUCKET_LOG_PATH, and BITBUCKET_DIFF_MAX_LINES_PER_FILE.
Restart the client to load the new server.

claude_desktop_config.json

{"mcpServers":{"bitbucket":{"command":"node","args":["/path/to/bitbucket-server/build/index.js"],"env":{"BITBUCKET_URL":"https://your-bitbucket-server.com","BITBUCKET_TOKEN":"your-access-token","BITBUCKET_DEFAULT_PROJECT":"your-default-project"}}}}

Generate and validate client config

Fit and risk

Best for

Teams running self-hosted Bitbucket Server / Data Center who want a conversational interface for PR workflows
CI/CD or proxy environments that need safe, read-only access to Bitbucket data
Organizations behind Zero Trust networks that need custom headers to reach Bitbucket

Not for

Users on Bitbucket Cloud (bitbucket.org) rather than Bitbucket Server/Data Center, since this server targets the Server API
Users unwilling to run a local Node.js MCP server process

Required permissions

Requires a Bitbucket Server personal access token, or username/password basic-auth credentials
The scope of allowed operations is bounded by the configured credential's repository permissions; write actions (create/merge/comment/delete branch, etc.) need corresponding write access
Optional custom HTTP headers are sent with every request and may carry sensitive tokens that must be managed carefully

Risks and side effects

Authentication tokens or passwords are configured as plaintext environment variables and must be stored/transmitted securely
Outside read-only mode, the model can directly merge, decline, delete branches, and delete comments — irreversible or collaboration-impacting actions with real misuse potential
The log file is written to local disk by default (`~/.bitbucket-server-mcp/bitbucket.log`) and may capture sensitive operation details

Troubleshooting

Verify `BITBUCKET_URL` and authentication (token or username/password) are correct and the account has the needed repository permissions
If write operations are rejected, check whether `BITBUCKET_READ_ONLY=true` was set unintentionally
If a tool call complains about a missing project, set `BITBUCKET_DEFAULT_PROJECT` or pass the `project` parameter explicitly
Check the log file (default `~/.bitbucket-server-mcp/bitbucket.log`, or the `BITBUCKET_LOG_PATH` override) for detailed error messages
Confirm `npm install` and `npm run build` were run, and that the client config path points to the correct `build/index.js`

Use cases

Create, review, approve, and merge Bitbucket Server pull requests directly from an AI assistant

Search code across projects, browse repository structure, and read file contents

Inspect pull request diffs, comments, review status, and CI/CD code insight results

Manage branches (list, filter, delete merged ones) and browse commit history

Supported clients

Claude DesktopFull support

VS CodeFull support

Related MCP servers

Compare with other MCP servers