Rollbar is an MCP server that connects AI tools directly to Rollbar project data for error monitoring, observability, and debugging. Built using the stdio transport, it enables developers and operations engineers to investigate production exceptions, inspect stack traces, and manage error tracking workflows without leaving their AI-assisted development environments. The server allows assistants to retrieve project lists, inspect full item details along with recent occurrences, and review deployment logs to correlate releases with bug spikes. Users can also assess code performance across specific versions or environments, surface the top active errors occurring over the past 24 hours, and fetch user session replays. In addition to read-only diagnostics, the server supports mutating item statuses, updating severities, adjusting assignments, and marking issues as resolved directly through chat. It can operate across single or multiple repositories using project-specific tokens or an account-wide access token.
Category: Developer Tools & Code Intelligence
Tags: debugging, error-tracking, logs, monitoring, rollbar
To install and configure Rollbar with an MCP client such as Claude Code, follow these steps: 1. Obtain a Rollbar project access token or account access token from your Rollbar settings. A read token suffices for data inspection, while read and write scopes are required for updating items. 2. Open your client configuration file, such as .mcp.json. 3. Add the server entry using the npx command: json { "mcpServers": { "rollbar": { "type": "stdio", "command": "npx", "args": ["-y", "@rollbar/mcp-server@latest"], "env": { "ROLLBAR_ACCESS_TOKEN": "<project read/write access token>" } } } } 4. For multiple projects, specify ROLLBAR_ACCOUNT_ACCESS_TOKEN in the environment block or create a .rollbar-mcp.json configuration file. 5. Save the configuration and restart your client to enable the tools.
Part of MCP Servers
The Rollbar MCP server connects Model Context Protocol clients to Rollbar error monitoring data. Running locally over stdio, it gives AI coding assistants programmatic access to review stack traces, analyze deployments, track error occurrence frequencies, and manage issue tracking states across single or multiple projects.
Read-only tools such as listing projects, reading occurrences, viewing deployments, and inspecting items require a token with read scope. If you want your AI assistant to run the update-item tool to change issue statuses or assignees, your token must have both read and write scopes enabled.
Any development tool or assistant supporting the standard Model Context Protocol stdio transport can use this server. This includes AI coding assistants like Claude Code and Cursor, which run the process directly without requiring a separate external web service or HTTP bridge.
Yes. You can provide a single account access token using ROLLBAR_ACCOUNT_ACCESS_TOKEN to access all projects in your account, or define a local .rollbar-mcp.json configuration file specifying separate credentials and tokens for each distinct project.