# LogRocket MCP Server An MCP server for integrating LogRocket data with your LLM applications ## Summary An MCP (Model Context Protocol) server that allows Cursor, Claude Code, Codex, and other MCP clients to query the LogRocket API for information about your LogRocket sessions, metrics, issues, and more. For example, allows you to: * Pull in context about your app's behavior to help with debugging and fixing errors * Get a better sense of how users interact with your app * Query a user or account's sessions for targeted investigation See [Ask Galileo](https://docs.logrocket.com/docs/ask-galileo) for information about what the LogRocket MCP server can help with. ## Example Prompts Here are some examples of how you can use the LogRocket MCP server in your AI-assisted workflow: * **Fix user-reported issues:** "User X reported a problem with checkout. Can you use LogRocket to watch their sessions, figure out the root cause, and fix it?" * **Understand feature usage:** "I'm about to work on the search feature — can you use LogRocket to help me understand how it's currently being used?" * **Triage new issues:** "Can you look at LogRocket for new issues from the past week, try to figure out their root causes, and then suggest which ones I can fix?" * **Check for regressions:** "Look at all commits from last week, and check LogRocket data to ensure they didn't introduce any regressions." * **Prioritize your work:** "Use LogRocket to watch sessions and look at issues to figure out what is highest priority that I work on next." ## Suggested Automations Because the MCP server can be called programmatically by AI agents, you can also set up powerful automations that continuously leverage LogRocket data: * **Research churning customers and low NPS scores:** Automatically pull LogRocket sessions for users who are churning or leaving low NPS scores to understand what went wrong in their experience. * **Research new support tickets:** Connect to your help desk to automatically research incoming support tickets using LogRocket session data. LogRocket offers out-of-the-box integrations with [Zendesk](https://docs.logrocket.com/docs/zendesk) and [Intercom](https://docs.logrocket.com/docs/intercom) to attach session replays directly to tickets. * **Summarize user behavior for sales and customer success:** Automatically generate summaries of how key accounts are using your product, giving your sales and customer success teams actionable insights. * **Connect LogRocket with your backend data:** Build a skill that allows your agent to correlate LogRocket frontend data with backend observability tools (e.g., Datadog MCP) for end-to-end debugging. * **Run daily or weekly reports:** Schedule an agent to look for new issues and UX frustration signals on a recurring basis, so your team is always aware of emerging problems. ## Setup The LogRocket MCP server is available at `https://mcp.logrocket.com/mcp`. The setup instructions below use this URL. It allows your MCP client to access the same LogRocket organizations and projects you can access in your browser. **Optional**: You can also choose to restrict your MCP client's access to a specific LogRocket organization or LogRocket project by adding to the MCP server URL's path: * To scope to an organization, use `https://mcp.logrocket.com/mcp/` * To scope to a project, use `https://mcp.logrocket.com/mcp//` You can find organization and project IDs in most LogRocket URLs (e.g., `https://app.logrocket.com///s/...`). You can also find them in the LogRocket dashboard under **Settings > Project Settings**. The App ID listed there has the form `/`. ### Tools By default, the LogRocket MCP server exposes 3 tools: * `list_organizations`: For listing the LogRocket organizations you can access. Hidden when scoping to a specific organization (e.g., `https://mcp.logrocket.com/mcp/`). * `list_projects`: For listing the LogRocket projects you can access within a given organization. Hidden when scoping to a specific project (e.g., `https://mcp.logrocket.com/mcp//`). * `use_logrocket`: For running natural language queries against your LogRocket data (e.g., sessions, metrics, and issues). Queries are handled by [Ask Galileo](https://docs.logrocket.com/docs/ask-galileo) which chains tool calls internally. You can access additional, specialized tools and customize which tools are exposed by using the `toolsets` query parameter when setting up your connection to the server. `toolsets` can be set to any comma-separated combination of supported values. For example, to access sessions and metrics tools, you'd use `?toolsets=sessions,metrics` (e.g., `https://mcp.logrocket.com/mcp?toolsets=sessions,metrics`, or if scoping to a project, `https://mcp.logrocket.com/mcp//?toolsets=sessions,metrics`). Supported toolsets and the tools they expose include: * `sessions`: Exposes a `find_sessions` tool for filtering LogRocket sessions and a `watch_sessions` tool for analyzing and/or extracting additional information about one or more of them. * `metrics`: Exposes a `build_metric` tool for querying LogRocket analytics data. * `ask-galileo`: Exposes the default `use_logrocket` tool. * `all`: Exposes all tools. Any new toolsets not currently listed here will be automatically enabled as they become available. For more control over which toolsets you're using, list them explicitly instead of using `all`. > NOTE: LogRocket MCP server tools are under active development and subject to change. Please contact or your CSM to share feedback or get help. ### Authentication Most MCP clients will ask you to authenticate using OAuth after you add the LogRocket MCP server. We recommend using OAuth when possible. It's the most secure authentication option we offer and will allow your MCP client to access the same LogRocket organizations and projects you can access in your browser. However, we understand that OAuth authentication can be difficult in some contexts, such as when running server-side. In these cases, you can use an API key to authenticate instead by sending the `Authorization` header on your MCP requests (e.g., `"Authorization": "Bearer "`). Some examples for popular MCP clients are provided below. API keys can be created in the LogRocket dashboard under **Settings > API Keys**. API keys are project-scoped, so you'll need to create one and configure a separate MCP connection for each project you want your MCP client to be able to access. ### Cursor [Install MCP Server](https://cursor.com/en-US/install-mcp?name=logrocket\&config=eyJ1cmwiOiJodHRwczovL21jcC5sb2dyb2NrZXQuY29tL21jcCJ9) To install manually: 1. Press `⌘` + `Shift` + `J` to open Cursor Settings. 2. Select **Tools & MCP**. 3. Select **New MCP Server**. 4. Add the following JSON: ```json { "mcpServers": { "logrocket": { "url": "https://mcp.logrocket.com/mcp" } } } ``` 5. **Connect** to the MCP server in the settings screen to authenticate. #### Using an API key Set the `Authorization` header to connect to the LogRocket MCP server with an API key instead of OAuth. Create an API key in the LogRocket dashboard under **Settings > API Keys**. Then configure the `Authorization` header as follows: ```json { "mcpServers": { "logrocket": { "url": "https://mcp.logrocket.com/mcp", "headers": { "Authorization": "Bearer " } } } } ``` ### Claude #### Claude Code 1. Open your terminal to access the CLI. 2. Run: ```bash claude mcp add --transport http logrocket https://mcp.logrocket.com/mcp ``` 3. Follow the prompts to authenticate. If authentication doesn't happen automatically, trigger it manually with `/mcp`. #### Claude Desktop 1. In the sidebar, select **Customize**. 2. In the new sidebar that appears, select **Connectors**. 3. Click the **+**, then select **Add custom connector** from the dropdown menu. 4. Enter a name (e.g., LogRocket) and LogRocket's remote MCP server URL, `https://mcp.logrocket.com/mcp`. 5. Click **Add**. 6. Click **Connect** under the newly-added LogRocket connector to authenticate. #### Using an API key (Claude Code) Set the `Authorization` header to connect to the LogRocket MCP server with an API key instead of OAuth. Create an API key in the LogRocket dashboard under **Settings > API Keys**. Then configure the `Authorization` header as follows: ```bash claude mcp add --transport http logrocket https://mcp.logrocket.com/mcp --header "Authorization: Bearer " ``` ### Codex 1. Open your terminal to access the CLI. 2. Run: ```bash codex mcp add logrocket --url https://mcp.logrocket.com/mcp ``` 3. Run: ```bash codex mcp login logrocket ``` And follow the prompts to authenticate. #### Using an API key Set the `--bearer-token-env-var` option to connect to the LogRocket MCP server with an API key instead of OAuth. Create an API key in the LogRocket dashboard under **Settings > API Keys**. Then export it as the value of an environment variable (e.g., `export LOGROCKET_API_KEY=`) and configure Codex as follows: ```bash codex mcp add logrocket --url https://mcp.logrocket.com/mcp --bearer-token-env-var LOGROCKET_API_KEY ``` ### VS Code [Install in VS Code](vscode:mcp/install?%7B%22name%22%3A%22LogRocket%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fmcp.logrocket.com%2Fmcp%22%7D) To install manually: 1. Press `⌘` + `P` and search for **MCP: Add Server**. 2. Select **HTTP (HTTP or Server-Sent Events)**. 3. Enter the URL `https://mcp.logrocket.com/mcp` and hit enter. 4. Enter the name **LogRocket** and hit enter. 5. Follow the prompts to authenticate. 6. Activate the server using **MCP: List Servers**, selecting **LogRocket**, then selecting **Start Server**. #### Using an API key Set the `Authorization` header to connect to the LogRocket MCP server with an API key instead of OAuth. Create an API key in the LogRocket dashboard under **Settings > API Keys**. Then go to **MCP: Open User Configuration** and configure the `Authorization` header in `mcp.json` as follows: ```json { "servers": { "logrocket": { "type": "http", "url": "https://mcp.logrocket.com/mcp", "headers": { "Authorization": "Bearer " } } } } ```