Everything bugstack maintains
Crashes, failed deploys, vulnerable packages, leaked secrets and outdated packages. Watched, fixed, and sent to you as pull requests. This page lists what is live today.
bugstack takes problems from whatever you already run.
import { ErrorCaptureClient } from 'bugstack-sdk'
ErrorCaptureClient.init({
apiKey: process.env.BUGSTACK_API_KEY,
endpoint: 'https://dashboard.bugstack.ai/api/ingest/sdk',
})Also for Python, Ruby and Go. The JavaScript SDK redacts API keys, tokens and passwords before an error leaves your app; in Python, Ruby and Go you strip anything sensitive with a before_send hook. Duplicate errors are grouped, so you see one row per real bug. Full setup in the Quickstart. More on data handling in Security.
Crash fixes
Your app reports uncaught server-side errors through the SDK. bugstack reads the failing code in your repository, works out the cause, and writes the fix. Anything over 3 files or 30 changed lines, or touching auth or security code, always waits for your review.
Failed-deploy fixes
Connect your host and bugstack hears about failed deploys. It works out what went wrong and opens a pull request with the fix. Render is the host supported today.
Dependency upkeep
bugstack checks your repository for outdated packages every day and reports what it finds. It can open pull requests for the upgrades. That part is switched on per account.
Security scanning
bugstack scans your repository for known-vulnerable packages every four hours and for leaked secrets every night, then reports what it finds. It can open pull requests for those findings, switched on per account, and a security upgrade always waits for you. bugstack never rotates a secret itself; it tells you.
Specialist agents
Each problem goes to an agent built for it. One diagnoses and patches. One reproduces the failure in the sandbox. Others handle security, dependencies and infrastructure. The fixes are written by Anthropic’s Claude.
Three autonomy levels
Set a level for each repository: Suggest, Review or Autopilot. New repositories start on Review, where bugstack opens the pull request and you merge it.
Proven in a sandbox
Every bug fix is proven in an isolated sandbox before you see it: a test fails on your current code for the bug’s own reason, then passes with the fix while the rest of your suite stays green. How it works ↓
Audit log and undo
Every decision bugstack makes lands on a tamper-evident audit log, so nothing happens off the record. If you want a change gone, undo opens a revert pull request.
Alerts where you work
Get notified by email, Slack, Microsoft Teams, PagerDuty, or in the app. Security changes always wait for a person, and so do infrastructure changes and rollbacks in production and staging.
Proven in a sandbox, before you see it
bugstack does not ask you to trust a fix. It proves it: in an isolated copy of your repository, a test reproduces the bug, then passes with the fix while the rest of your suite stays green.
Copied in, sealed off
Your repository is copied into an isolated sandbox. Dependencies install from your package registries, then the network is cut. None of bugstack’s keys and none of your secrets go in.
The bug, reproduced
The tester reads the code the error comes from and your nearest tests, and writes a test that fails on your current code, for the bug’s own reason. A failure for any other reason, a typo or a bad import, never counts.
The fix, proven
The fixer writes the change in the same sandbox. The test has to pass and every other test has to stay green. If not, it tries again with what the sandbox printed. A fix that doesn’t pass never becomes a pull request.
The proof, in the pull request
The pull request carries the test, the error it failed with before the change, and the tests still passing. At merge, bugstack runs the proof again, and Autopilot merges only a proven fix.
Runs Python (pytest), JavaScript and TypeScript (Vitest, Jest, Mocha) and Go test suites. Then the sandbox is destroyed. How it is isolated →
Works with the tools you already use
Four SDKs, GitHub pull requests, and Claude writing the fixes
Beyond monitoring
Monitoring tells you something broke. bugstack does the work.
Tools like Sentry, Datadog and PagerDuty tell you when something goes wrong. bugstack picks up from there: it finds the cause, writes the change, and opens a pull request. It also watches your deploys and the packages in your repository. It does not replace dashboards, tracing or performance monitoring.
| Capability | Error Monitoring | bugstack |
|---|---|---|
| Detects production errors | ✓ | ✓ |
| Sends alerts & notifications | ✓ | ✓ |
| Dashboards, tracing and performance monitoring | ✓ | ✕ |
| Finds the cause in your code | Partial | ✓ |
| Writes the fix | ✕ | ✓ |
| Opens a pull request | ✕ | ✓ |
| Runs the fix through your CI | ✕ | ✓ |
| Fixes failed deploys (Render today) | ✕ | ✓ |
| Scans your repository for vulnerable packages, secrets and outdated packages | ✕ | ✓ |
| Lets you set how much happens without you | ✕ | ✓ |
| Proves each fix in an isolated sandbox before you see it | ✕ | ✓ |
| Merges on its own only with sandbox proof | ✕ | ✓ |
| Logs every decision, with undo | ✕ | ✓ |
bugstack works without a monitoring vendor. If you have one, keep it. See the full comparison →
Hand off the upkeep. Keep building.
Every change is a pull request. You decide how much happens without you.
Get started