Check’d
Start free

Product

All featuresProduction monitoringSession replayAI explain and fixImport API docsJourneys
DevelopersPricingFree site scanChangelogSign in

From a stack trace to a fix you can review.

Press Explain with AI on any issue. Check’d reads the evidence, never your users’ personal data, and your code at the matching release, then says what is broken, why, where to look and how to fix it. Happy with it? Open a fix pull request.

Start freeHow it works in detail

Explanations are on every plan and use your usage. Fix pull requests need GitHub on a paid plan.

An answer in six parts

Written for whoever picks the issue up, not only the person who wrote the code.

  • What is broken

    The problem in plain words, the way you would explain it to a teammate.

  • Likely cause

    The root cause, read from the error, your frames and the steps before it.

  • Who is affected

    How many people and how badly, from the occurrences and users counted on the issue.

  • Where to look

    Files, lines and functions, each with why it matters.

  • Suggested fix

    The change to make, step by step.

  • How sure

    High, medium or low confidence with the reason, and what it could not know.

What the AI is given, and what it never sees

The context is built from the issue’s evidence, with personal data left out before anything is sent.

Given

  • The error type, with numbers, ids, quoted values and URLs taken out of the message
  • The stack frames in your own code
  • Breadcrumbs as page routes
  • The failed request as method, host, route and status
  • The spread of browsers, systems and releases, and how many people and occurrences
  • Linked tests, and with GitHub connected, the lines around your frames

Never given

  • User ids, emails or their hashes
  • Anonymous and session ids
  • IP addresses
  • Cookies and tokens
  • Query strings
  • Request and response bodies
  • Tags and extra data

Open a fix pull request from the issue

With GitHub connected and pull requests allowed for the project, Check’d drafts a change from the issue, its explanation and the files in its stack, on a new branch.
  • You review every diff before anything is pushed. Nothing is merged for you.
  • The pull request links back to the issue, and the issue shows its pull request.
  • Once the fix is merged and a later release shows no new occurrences, the issue suggests resolving it.
  • The same flow works from findings in test runs and scans.

How it fits together

  1. Step 1: An issue arrives

    From production monitoring, a test run or a scan.

  2. Step 2: Explain with AI

    The answer stays on the issue and offers a refresh when newer occurrences or a newer release arrive.

  3. Step 3: Open a fix pull request

    Review the diff on GitHub and merge it yourself.

  4. Step 4: Confirm the fix

    Re-run the failed tests, or watch the next release stay clean.

Questions people ask

What the AI reads, what it never reads, and what it costs.

Ask support something else
What does Explain with AI tell me?

What is broken in plain words, the likely root cause, who is affected and how badly, where in your code to look, a suggested fix, and how sure it is, with the reason and anything it could not know.

Which code does it read?

When the project is connected to GitHub, it reads the lines around the stack frames in your own code, at the commit or tag of the release when one matches. Otherwise it reads the default branch and says so.

Is my users’ data sent to the AI?

No personal data. User ids, emails, anonymous and session ids, IP addresses, cookies, tokens, query strings, bodies, tags and extra data are never given to the AI. Numbers, ids, quoted values and URLs are taken out of error messages first.

Does it change my code?

Not by itself. Open a fix pull request drafts a change on a new branch, you review every diff before anything is pushed, and nothing is merged for you.

What does it cost?

Each explanation counts toward your plan’s usage like other AI work. Fix pull requests need GitHub, which is on the paid plans, and share the plan’s monthly fix allowance: 5 on Starter, 40 on Business and unlimited on Team.

Does it work on test findings too?

Yes. Findings from test runs and scans already carry the explanation their run wrote, and the same fix pull request flow works from them.

Works with the rest of Check’d

Get your first explanation today

Explain with AI is on every plan. Fix pull requests start on Starter.

Start freeScan a site, no account needed