Flare AI
Beta
It finds the cause of errors and slow routes, then opens the fix. Use the Flare Assistant, or your own agent with the Flare MCP server or CLI. Included in every plan, at no extra cost.
The Flare Assistant and coding agents connected through the Flare MCP server or CLI take the same steps.
Works in
Both see the same errors, logs, and traces.
Flare Assistant
Open the Flare Assistant next to any error, log, or slow route. It knows what you’re looking at and reads your GitHub repo to find the cause.
Pull requests need the GitHub integration.
Your coding agent
Claude Code, Codex, Cursor, or the agent in your terminal. Connect the Flare MCP server or CLI, and your agent gets the same context next to your local checkout.
Not just errors. Ask about slow routes, queries, jobs, and logs too.
Errors
Stack traces, occurrences, affected users, request context, and spikes.
“Why did this error spike after the last deploy?”
Logs
Search and filter by message, severity, host, and entry point type.
“Are there related logs around the first occurrence?”
Routes
Response times, trends, and the slowest traces for every route.
“Which routes got slower this week?”
Queries
The database queries behind a slow route, job, or command.
“Which queries are behind the slow checkout?”
Jobs & commands
Queued jobs and Artisan commands: duration, failures, and slowest runs.
“Which jobs take longer than usual?”
HTTP calls & Livewire
External API calls and Livewire components, with their slowest traces.
“Which external API is slowing down checkout?”
Learn more about error tracking, performance monitoring, and logging.
An MCP server for your Laravel & PHP apps, or the CLI. You only need one.
Add Flare as a remote MCP server. Your client opens Flare in your browser, you pick the teams and projects, and you’re connected. No API keys to paste.
claude mcp add \
--transport http \
flare https://flareapp.io/mcpInstall the CLI, log in through your browser, and add the Flare skill to your project. The skill teaches your agent the Flare commands.
composer global require spatie/flare-cli
flare login
flare install-skillNo connection yet? Copy any error as Markdown, with its details and the first 20 stack frames. Or copy a ready-made MCP or CLI prompt with the IDs filled in.
Please use the Flare CLI to investigate this error and suggest a fix. Error: Illuminate\Database\Eloquent\MissingAttributeException: The attribute [lifeboat] either does not exist or was not retrieved for model [App\Models\Passenger]. Error ID: 48213 Project ID: 1912 Project: Titanic Team: White Star Line Flare URL: https://flareapp.io/errors/48213-the-attribute-lifeboat-either-does-not-exist-or-was-not-retrieved-for-model-appmodelspassenger/latest Use the Flare CLI to fetch the latest occurrence, inspect the stack trace and context, then suggest the code change that should fix it.
Building your own integration? The same data is available through the REST API for scripts, CI jobs, and internal tools.
curl https://flareapp.io/api/projects \
-H 'Authorization: Bearer <token>' \
-H 'Accept: application/json'AI does the digging. You still hit merge.
5 million AI tokens per team each month, trials too. No seat fees, no usage bills.
One switch for team admins. Flare then sends no error, log, or performance data to OpenAI.
The Flare Assistant asks first. Code changes go to a pull request, a draft by default.
GitHub access needs a team admin. MCP and CLI connections only see the projects you pick.
Paste one prompt into your coding agent. It installs and configures Flare.
Use the Flare Assistant when you’re already looking at the problem in Flare. Use your own agent when the fix needs your local setup or tests.
Only through a pull request you approve. It prepares the exact change and shows you the diff. After you approve, it pushes the change to its own branch and opens a pull request, a draft by default. It never merges, force-pushes, or writes to your default branch.
The Flare Assistant, AI-written issues, and docs search use OpenAI. Team admins can turn off AI features at any time. Flare then sends no error, log, or performance data to OpenAI. Docs search queries still go to OpenAI. The MCP server and CLI work with whichever AI tool you choose.
Yes. AI is included on all Flare plans, with no per-seat AI add-on. Every team gets 5 million AI tokens per month for the Flare Assistant and AI-written issues, including during the trial. Usage stops at that limit, so there’s never an extra bill. The MCP server and CLI don’t use these tokens.
MCP (Model Context Protocol) is a standard way for AI tools to connect to other apps. The Flare MCP server at https://flareapp.io/mcp lets Claude Code, Cursor, Codex, and other clients read your errors, logs, and performance data. With your permission, they can also resolve, snooze, and add notes to errors. Follow the setup guide for your client.
No. Choose MCP for a direct connection from your AI tool, or the CLI if your agent works through terminal commands. You can also copy an error as Markdown without setting up a connection.
Let the Flare Assistant open the fix, or connect your own agent in a minute.
10-day trial · No credit card required
Laravel Developer
“I had already fixed the bug in Flare before my colleagues came to me to report the error. Nice work!”
CTO at Ofelos
“Flare allows us to catch, understand & fix errors in minutes. Setup for a new project takes only seconds. It increases the quality of our product AND saves us time. Awesome stuff!”