Skip to main content
Checksum solves this with two complementary systems that keep your test suite green automatically:
  1. Auto-Recovery — Real-time fix attempts during test execution
  2. Auto-Healing — Batch repair of failing tests after CI runs, delivered as PRs
Together, they create a zero-maintenance testing experience.
Auto-Maintenance Overview

​
Why Tests Break

Common reasons tests fail after app changes:
CauseExample
Selector driftA button’s data-testid or text changed
Timing issuesA page loads slower after a new feature was added
Assertion mismatchesThe expected text or value changed
Flow changesA step was added or removed from a user flow
Setup driftTest data requirements changed
Checksum handles all of these automatically.

​
The Two Layers

​
Layer 1: Auto-Recovery (During Execution)

When a test step fails during execution, the CLI’s AI fallback kicks in immediately and attempts to recover on the fly. This happens transparently — if recovery succeeds, the test passes and the healed count increments. → Learn more about Auto-Recovery

​
Layer 2: Auto-Healing (After CI Runs)

When tests fail in CI and can’t be auto-recovered, Checksum creates a healing session — an AI agent that analyzes the failures, fixes the test code, and opens a PR with the fixes. This supports both Standard and Deep modes. → Learn more about Auto-Healing

​
The Maintenance Loop

Tests run in CI
  → Some tests fail
    → Auto-recovery fixes them in real time (if possible)
    → Remaining failures trigger auto-healing
      → AI agent fixes the tests
      → PR opened with healed tests
        → You review and merge
          → Tests are green again

​
Next Steps