Know about crashes before your users tell you.

BugFixes is error reporting for your apps. Add the library to your TypeScript, Go or Rust code and every unhandled error is caught, matched against errors you have already seen, and turned into a ticket in the tracker your team already uses.

Your team gets an alert in Slack, Discord, Microsoft Teams or Google Chat, with the error, stack trace and ticket link ready to act on.

Every account starts on the Free plan — 25 AI analyses and 50 tickets a month, no card needed. See what's included

What it looks like when it happens

No demo data. This is a real crash from our own dashboard on 30 September 2026, exactly as the system handled it — from the first report to the suggested patch, in under 15 seconds.

1. 1.A crash came in

   19:46:36 UTC — the TypeScript agent in a Next.js API route reported it.

   Error

   TypeError in validateEnvironmentResponse

   Where

   app/api/environment/create/route.ts:40

   Fingerprint

   New — first occurrence. Repeats of this crash will update the same ticket instead of opening another.

2. 2.A ticket was opened in the linked GitHub repository

   19:46:37 UTC — issue #2068, written from the deterministic analysis, no AI involved yet.

   TypeError in validateEnvironmentResponse

   TypeError originated in app/api/environment/create/route.ts at line 40 inside validateEnvironmentResponse. The program likely tried to use a value that was missing or had an unexpected type.

   • Type: TypeError
   • Occurrences: 1
   • Location: app/api/environment/create/route.ts:40
3. 3.The team was told in Discord

   Seconds later — one message, linking to the ticket. Repeats of the same crash do not spam the channel.

   Bugfixes report: TypeError in validateEnvironmentResponse

   TypeError originated in app/api/environment/create/route.ts at line 40 inside validateEnvironmentResponse.

   Ticket: 2068

   github.com/flags-gg/dashboard/issues/2068

4. 4.AI analysis landed on the ticket

   19:46:51 UTC — BugFixes AI read the linked repository and commented with a diagnosis, how to validate it, and a patch.

   Diagnosis. The stack trace references validateEnvironmentResponse at line 40, but the current source does not contain that function — a revision mismatch. The deployed code validates the environment-creation response and throws when environment_id is missing or not a non-empty string.

   Validation. npm run lint; npm run test, then POST to /api/environment/create. Risks: callers that received the raw upstream payload will now get a 500 when environment_id is missing; check the logger does not expose response data.

   The AI also said, plainly, that the diff would not apply to the checked-out revision — so the developer knows before trying.

diff --git a/src/app/api/environment/create/route.ts b/src/app/api/environment/create/route.ts
--- a/src/app/api/environment/create/route.ts
+++ b/src/app/api/environment/create/route.ts
@@
   const body = await response.json()
-
-    return NextResponse.json(body)
+    // Validate the shape of the response before forwarding it.
+    // The downstream client expects `environment_id` to be a non-empty string.
+    if (!body || typeof body.environment_id !== 'string' || body.environment_id.trim() === '') {
+      logError('Created environment response failed schema validation', new Error('expected environment_id to be a non-empty string'))
+      return NextResponse.json({ error: 'Invalid environment creation response' }, { status: 500 })
+    }
+
+    return NextResponse.json(body)

What you get

• Repeat crashes bump the priority

  When the same error keeps happening, the existing ticket is raised in priority instead of your tracker filling with duplicates. The bugs hurting most rise to the top on their own.
• No log trawling

  The ticket already holds the error, the stack trace and the file and line. Nobody has to dig through logs to work out what happened.
• No waiting on a user to complain

  You hear about a crash the moment it happens, not days later when someone gets round to emailing support.
• Ask your AI assistant about your crashes

  BugFixes has an MCP server, so Claude Code, Codex, Cursor, VS Code and Gemini CLI can list your projects, read crash reports and update their status, all from your editor. Connect with a personal access token from your settings.
• Your code and crash data stay yours

  Use BugFixes AI for a generated patch, or bring your own AI when your source code must stay in your environment. BugFixes stores only a crash checksum to recognise repeats, then removes the report after sending it to your ticketing system.

How it works

1. 1.Your app throws

   The library catches the error or crash and sends the message, stack trace and where it happened to BugFixes. Nothing to wire up beyond an agent key.
2. 2.We check if we have seen it

   Every report is fingerprinted. If the same error has come in before, it is treated as the same bug rather than a new one.
3. 3.A ticket is raised

   A new bug gets a ticket in Tracklines, Linear, Jira or GitHub Issues with the error, stack and file and line already filled in. Nobody has to write it up.
4. 4.Your team is told

   A message goes to your Slack, Discord, Microsoft Teams or Google Chat channel saying what broke, with a link to the ticket.
5. 5.A fix is suggested

   Use BugFixes AI to generate a patch, or connect your own AI and keep your source code in your environment. Developers stay in control of the implementation.

Simple pricing, no surprises

Every account starts on the Free plan — no card needed. Upgrade when you want unlimited tickets or BugFixes-managed AI, for one flat price per team.

Full plan details, FAQ and how we compare with Sentry, Rollbar, BugSnag, Raygun and Airbrake