Skip to main content

Bitrise MCP

Abstract

The Bitrise Model Context Protocol (MCP) Server lets you talk to Bitrise via an AI client of your choice. It enables seamless interaction with your existing CI setup, including troubleshooting and automation.

The Bitrise Model Context Protocol (MCP) Server lets you talk to Bitrise via an AI client of your choice. It enables seamless interaction with your existing CI setup:

  • Troubleshoot issues by directly asking about failure reasons. The AI analyzes logs and configurations, providing actionble recommendations.

  • Optimize configurations by instructing the AI to suggest improvements. You will receive instant insights and practical suggestions.

  • Automate manual tasks such as handling permissions, inviting members to a project, or looking for old builds. Tell the AI agent about your requirements: it will send you the proposed steps for review and ask for permission before executing each action. You can review the results at each step of the process.

For now, the Bitrise MCP is supported for the Claude desktop app. You can find the Bitrise MCP repository on GitHub.

Configuring the Bitrise MCP with Claude

The Bitrise MCP supports the Claude desktop app. To set it up:

  1. Generate and copy a Bitrise personal access token.

  2. Make sure you have uv installed on your device.

  3. Download the Claude desktop app.

  4. In the app, generate a claude_desktop_config.json file: from the Claude menu, select Settings and then Developer, then click Edit config.

  5. Add the following content to the JSON configuration file:

    Remove placeholders

    Make sure to replace:

    • <ABSOLUTE_PATH_TO/uvx> with the actual path to uvx on your device.

    • <YOUR_ACCESS_TOKEN> with your Bitrise personal access token.

    {
      "mcpServers": {
        "bitrise": {
          "command": "<ABSOLUTE_PATH_TO>/uvx",
          "env": {
            "BITRISE_TOKEN": "<YOUR_ACCESS_TOKEN>"
          },
          "args": [
            "--from",
            "git+https://github.com/bitrise-io/[email protected]",
            "bitrise-mcp"
          ]
        }
      }
    }
  6. Save the file and refresh the Claude app. You should have all Bitrise tools available in your chat with the Claude AI agent.

    Click the hammer icon to bring up the list of available tools. You can also check them at the Bitrise MCP repository.

  7. Optionally, you can limit the number of available tools:

    {
      "mcpServers": {
        "bitrise": {
          "command": "<ABSOLUTE_PATH_TO>/uvx",
          "env": {
            "BITRISE_TOKEN": "<YOUR_ACCESS_TOKEN>"
          },
          "args": [
            "--from",
            "git+https://github.com/bitrise-io/[email protected]",
            "bitrise-mcp",
            "--enabled-api-groups",
            "cache-items,pipelines"
          ]
        }
      }
    }

    In this example, Claude will only have access to the cache-items and pipelines tools.