Find the root cause.Review the fix as a PR.

Mendry is a self-hosted AI tool for production incident investigation. It connects alerts, logs, and code to find the root cause. With automatic hotfix enabled, it generates a patch within the limits you set and opens a draft PR for your team to review.

Install with AI (source evaluation)

Source preview · Product status
Mendry / incidentILLUSTRATIVE EXAMPLE

Production signal / checkout-api

Checkout requests are failing

INC / 0042

01 / Signal

Checkout errors increase.

A production alert and runtime logs point to the checkout service.

02 / Investigation

ERROR checkout-api

TypeError: session is nullcheckout.ts:4242 session.user.id

The code accesses a missing session.

Correlate the exception with repository context and retain the evidence behind the diagnosis.

03 / Patch

CANDIDATE DIFF

+ if (!session) {+   return unauthorized();+ }  const id = session.user.id;

Guard the expired-session path.

Automatic hotfix generates a candidate patch within the project's allowed paths and change limits.

  1. Local prevalidation

    Optional · isolated container tests

  2. Review delivery

    Review branch / Draft PR or MR

  3. Result notification

    Telegram, Feishu, or WeCom

  4. Team review

    Repository CI, merge & deployment

Illustrative automatic-hotfix flow, not a live run. Local prevalidation is optional; delivery depends on repository support and credentials.
  • Evidence-linked diagnosis
  • Policy-constrained patches
  • Reviewable Git delivery

Investigate. Repair. Hand off.

Start with analysis only. Enable automatic hotfix per project to carry a diagnosis through to a review branch or draft change request.

INVESTIGATION · DEFAULT

Analysis only
  1. Collect signals

    Ingest alerts and runtime observations in the context of a project.

  2. Investigate root cause

    Connect logs, stack traces, and the deployed code baseline.

  3. Review findings

    Inspect the diagnosis, evidence, recommended plan, and candidate diff.

Enable per project · Configure policy & write credentials

AUTOMATIC HOTFIX · OPT-IN

Available in preview
  1. Generate a bounded patch

    Restrict edits by allowed paths, protected files, and change size.

  2. Prevalidate locally · optional

    With enhanced validation configured, run baseline and patched tests in a network-isolated container.

  3. Deliver for review

    Push a review branch and create a Draft PR/MR where the repository adapter supports it.

Your team takes it from here

  1. Repository CI
  2. Review & merge
  3. Deploy
  4. Confirm recovery

Mendry does not automatically merge, deploy, roll back, or confirm production recovery. Operators update the incident status after checking the result.

Analysis only is the default. AWS-source incidents remain analysis only.

Product status

One workspace for the incident

Follow incoming signals, prioritize open incidents, and inspect each investigation and repair attempt.

Connect evidence to delivery

Bring signed webhooks, supported log sources, and a Git baseline into the same workflow. Configure repair policy and repository access for each project.

Explore automatic hotfix
  • Alerts
  • Logs
  • Git baseline
MendryInvestigate → patch → review

DELIVERY FOR YOUR TEAM

  • Review branch
  • Draft PR / MR
  • Self-hosted

    Run the service and configure its evidence sources in your own environment.

  • Bounded changes

    Define where repairs can make changes and how large a patch may be.

  • Traceable attempts

    Inspect execution records, validation results, and delivery details in the incident.

Choose your next step