Skip to main content
You connect your app once. Squasher then groups its errors into issues, and AI triage writes the likely cause, the impact, and a next step for each new issue.

​
What you can do

  • Catch errors. Send errors with an SDK, a log drain from your host, or OpenTelemetry. Squasher groups the repeats of one error into one issue.
  • Find the cause. Read the AI triage summary on the issue. Search the logs and traces around the error in Logs and Traces, and watch session replays from the browser.
  • Fix it. Fix the code yourself, or select Create fix on the issue, and Squasher opens a pull request on GitHub. Auto-fix needs the Team plan or above.
  • Get alerted. Monitors check your endpoints and open incidents. Notifications send them to Slack, PagerDuty, email, Telegram, or a webhook.
  • Tell your users. Publish a status page from your monitors and incidents.
  • Watch your AI features. Record model calls, tool calls, tokens, and cost from your AI code.

​
How it works

1

Your app sends data

An SDK, a log drain, or an OpenTelemetry exporter sends errors, logs, and traces to your Squasher project.
2

Errors become issues

Squasher groups the repeats of one error into one issue. The issue counts the events and shows when the error was first and last seen.
3

AI triage reads each new issue

For a new issue in production, AI triage writes the likely cause, the impact, and a next step. It reads the issue, its events and stack traces, and the project’s logs, traces, metrics, and deployments. It can only read, and it does not read your code.
4

Your team responds

Notifications tell the right people. You fix the code, or auto-fix opens a pull request for you to review.

​
Where to start

Quickstart

Connect your app and see your first issue.

Set up with a coding agent

Give your coding agent one prompt, and it installs Squasher.

Choose how to send data

Compare SDKs, log drains, and OpenTelemetry.

Use Squasher from an AI agent

Read these docs as Markdown, and use the MCP server, CLI, or API.