Our MCP server

The Flare MCP server allows you to integrate Flare's error tracking, performance monitoring, and logging capabilities directly into your AI-powered development workflows. Use it with AI assistants like Claude to automatically monitor, analyze, and resolve application errors, investigate performance issues, and search your logs.

See what you can do with Flare and AI.

What is MCP?

MCP (Model Context Protocol) is an open protocol that enables AI assistants to securely connect with external data sources and tools. By using the Flare MCP server, your AI assistant gains direct access to your error tracking, performance monitoring, and logging data and can perform actions in Flare on your behalf.

Key features

With the Flare MCP server, you can ask your AI assistant to:

"Show me the latest errors from my project"
"Fix the error regarding the PHP syntax error in myfile.php"
"Mark the error as resolved in Flare"
"Fix error ERROR-ID-HERE and mark the error as resolved"
"Show me all slow routes in my app"
"Which users are affected by this error?"
"List the slowest database queries from the last 7 days"
"Show me the slowest traces for this route"
"List the slowest external HTTP calls from the last 7 days"
"Snooze this error until next week"
"Assign this error to Alex"
"Create a note in Flare with the results of your investigation about this error"
"Show me the latest logs from my project"
"Search the logs for any database connection errors"
"Create a Flare project and use the documentation to install Flare in my application"

Authentication

The Flare MCP server is at https://flareapp.io/mcp and authenticates with OAuth 2.1. Every supported client opens a browser, sends you to Flare to grant access, and stores a refreshable token. No tokens are pasted by hand and the client never sees your password.

When you authorize, you pick which teams and projects the connection can access — the same scope picker used for personal access tokens. Grant the admin scope if you want your assistant to create projects.

If your client doesn't support MCP OAuth yet, you can fall back to a personal access token sent as a Bearer header.

Installation

Pick your client and follow the steps. After installation, jump to Project hint so your assistant knows which Flare project to use.

Amp

Add the server with the Amp CLI:

amp mcp add flare https://flareapp.io/mcp

You can also add the server to your Amp user settings in ~/.config/amp/settings.json:

{
  "amp.mcpServers": {
    "flare": {
      "url": "https://flareapp.io/mcp"
    }
  }
}

When you start Amp again, it opens your browser so that you can authorize Flare. Amp stores the token and refreshes it when necessary. If you must authorize again, run amp mcp oauth logout flare and restart Amp.

Amp orbs cannot open a browser for OAuth. To use Flare in an orb, follow Amp's MCP documentation and send a personal access token in the headers of the server.

Claude Code

Add the server with the Claude Code CLI:

claude mcp add --transport http flare https://flareapp.io/mcp

Then in a Claude Code session, run /mcp and follow the browser prompt to authorize. Claude Code stores the token. If it asks you to authorize again, run /mcp and approve in the browser.

Verify with:

claude mcp get flare

Claude Desktop and claude.ai

Custom connectors live in the same place on Claude Desktop and claude.ai:

Open Customize → Connectors.
Click + Add, then Add custom connector.
Name the connector Flare and set the Remote MCP server URL to https://flareapp.io/mcp.
Click Add, then complete the OAuth flow in your browser.

On Team and Enterprise plans, an owner adds the connector once under Organization settings → Connectors. Members then find Flare under Customize → Connectors and click Connect to authorize.

Codex (CLI, desktop app, and IDE extension)

The Codex CLI, desktop app, and IDE extension share a single config file at ~/.codex/config.toml. Add:

[mcp_servers.flare]
url = "https://flareapp.io/mcp"
oauth_resource = "https://flareapp.io/mcp"

Then start the OAuth flow:

codex mcp login flare

If you set up the server through the desktop app's Settings → MCP panel instead, the same entry appears in config.toml and the OAuth flow runs in your browser when you first use the server.

Cursor

Add Flare in Cursor Settings → MCP → Add custom server, or drop this into .cursor/mcp.json (project) or ~/.cursor/mcp.json (global):

{
  "mcpServers": {
    "flare": {
      "url": "https://flareapp.io/mcp"
    }
  }
}

Cursor opens a browser to authorize the first time the server is used and stores the resulting token for you.

OpenCode

Add Flare to opencode.json (project root) or ~/.config/opencode/opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "flare": {
      "type": "remote",
      "url": "https://flareapp.io/mcp"
    }
  }
}

Then authorize:

opencode mcp auth flare

OpenCode discovers the OAuth endpoints automatically via dynamic client registration (RFC 7591) and stores tokens at ~/.local/share/opencode/mcp-auth.json.

VS Code

Add Flare to .vscode/mcp.json (project). For a global setup, run MCP: Open User Configuration from the Command Palette and add the same entry there:

{
  "servers": {
    "flare": {
      "type": "http",
      "url": "https://flareapp.io/mcp"
    }
  }
}

Then run MCP: List Servers, select flare, and choose Start Server. VS Code opens a browser to authorize and stores the resulting token for you.

Alternative: pinned access token

If your client doesn't support MCP OAuth yet, create a personal access token with read and write scope and send it as a Bearer header. Add the admin scope if you want your assistant to create projects.

If your client can send custom headers, set Authorization: Bearer <YOUR-PERSONAL-ACCESS-TOKEN> on the server entry. If it only runs local servers, use mcp-remote as a bridge:

{
  "mcpServers": {
    "flare": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://flareapp.io/mcp",
        "--header",
        "Authorization:${AUTH_HEADER}"
      ],
      "env": {
        "AUTH_HEADER": "Bearer <YOUR-PERSONAL-ACCESS-TOKEN>"
      }
    }
  }
}

Project hint

To help your AI assistant pick the right project when there's more than one, add a one-line hint to your project's instructions file (for example CLAUDE.md, AGENTS.md, or your IDE's equivalent):

When the user asks to show or investigate errors, performance issues,
or logs, use the Flare MCP server. The project name in Flare is
INSERT-YOUR-PROJECT-NAME-HERE.

Alternative: agent skill

If your agent works through terminal commands, you can use the Flare CLI with the Flare agent skill. The skill adds instructions to your project that teach compatible coding agents how to investigate with the CLI. Choose MCP or the CLI based on how you want your agent to connect to Flare.

Available tools

Once configured, your AI assistant can:

List projects: View all projects associated with your Flare account, or find a project by name
Show errors: Display recent errors from a specific project
View error details: Get full information about a specific error including stack traces and context
Resolve errors: Mark errors as resolved directly from your AI conversation, or as unresolved again
Snooze errors: Snooze errors to temporarily suppress notifications (until a date, forever, for a number of occurrences, or until the next application version)
Unsnooze errors: Resume notifications for previously snoozed errors
Assign errors: Assign errors to team members for follow-up
List team members: View team members for a project to find assignees
Search errors: Find errors matching specific criteria
View error insights: See how often an error occurred, grouped by user, URL, job, version, type, or command
Create notes: Add investigation notes to errors for team collaboration
Browse logs: Search, filter, and paginate through project logs by severity, hostname, entry point type, or free-text search
Monitor performance: List the slowest routes, queries, jobs, commands, external HTTP calls, views, Livewire components, and browser pages and components, then inspect detailed metrics, trends, and the slowest traces for any of them
Create projects: Create a new Flare project in one of your teams and get the API keys to install Flare
Read documentation: Search and read the Flare documentation, so your assistant can install and configure Flare for you

Security

Each OAuth connection from a client to the Flare MCP server has its own name and is scoped to the permissions, teams, and projects you select during the consent flow. Multiple connections from Claude, Codex, Cursor, or other clients can work at the same time. Rename, adjust, or revoke one connection without affecting the others from Account → Connections. Personal access tokens used as a Bearer fallback are managed under Account → Personal access tokens.

Feedback

If you have any feedback or questions, please reach out to us at [email protected].

Getting started Agent skill

On this page

What is MCP?
Key features
Authentication
Installation
Alternative: pinned access token
Project hint
Alternative: agent skill
Available tools
Security
Feedback