Zero-Breakage Guarantee
Tests fail? We report, never PR. Your main branch stays safe — every update is verified before it reaches you.
Deep Dive
The guarantee that changes everything
Other tools open PRs and hope for the best. Ovvoc never opens a PR unless tests pass. If something breaks, you get a detailed failure report instead.
This is the fundamental difference. Your main branch stays safe. Your CI stays green. You never have to deal with a broken dependency PR.
The failure report isn't just "tests failed." It includes the exact error output, which transform was applied, what file and line caused the issue, and a suggestion for how to fix it manually.
Tests Pass
PR created with confidence score, full changelog, and test summary.
Tests Fail
Detailed failure report with error output, transform details, and fix suggestions. No PR created.
How It Works
Five stages, two possible outcomes
Clone
Your repository is cloned into a fresh, ephemeral container. Completely isolated from other jobs.
Update & Transform
Dependencies updated, AST transforms applied. The code is modified to match the new API surface.
Build
npm run build executes in the container. If the build fails, the pipeline aborts and generates a failure report.
Test
npm test runs your full test suite. If any test fails, you get a detailed report. No PR is created.
PR or Report
Both pass? A PR is created with a confidence score and full changelog. Either fails? A failure report with exact errors.
Technical Demo
When a test fails
Watch Ovvoc detect a test failure, abort the PR, and generate a detailed failure report.
Use Cases
Reports that actually help
Integration Test Catches Type Mismatch
A breaking type change in @types/express causes 3 integration tests to fail. The report identifies exactly which tests, which types, and suggests the fix.
Multi-Package Version Conflict
Updating React without updating react-dom creates a version mismatch. The report identifies the incompatibility and recommends an atomic update instead.
Build Failure from Removed API
An SDK removes a deprecated method your code uses. The build fails, and the report shows the exact call site, the removed method, and its replacement.
Testing Methodology
Baseline-aware test comparison
Baseline Capture
Before any changes, Ovvoc runs your test suite to establish a baseline. Tests that already fail are recorded. This is your starting point — the ground truth of your current codebase.
Post-Transform Run
After applying dependency updates and code transforms, the full test suite runs again. Every test result is captured and compared against the baseline.
Regression Detection
Only new failures count as regressions. If a test was already failing before the update, it is tracked but not blamed on the update. This prevents false positives from pre-existing issues.
The Guarantee
What “zero breakage” actually means
“Zero breakage” means zero regressions introduced by Ovvoc. It does not mean your codebase is bug-free. It means that when Ovvoc opens a PR, every test that was passing before the update is still passing after the update.
If any test that was previously passing now fails after the update, Ovvoc does not create the PR. Instead, you receive a detailed failure report explaining exactly what broke, which transform was applied, which file and line caused the issue, and a suggested path to resolution.
This is a fundamentally different approach from tools that open PRs and let CI catch failures after the fact. With Ovvoc, your main branch stays clean. Your CI stays green. Your team is never disrupted by a broken dependency PR that someone has to triage and close.
The guarantee is simple: if Ovvoc opens a PR, it is safe to merge. If it is not safe, you get a report instead.
Build System Support
Your build system, your rules
Ovvoc respects your project's build configuration. The build script from your package.json is executed exactly as your CI would run it. Whether you use webpack, vite, rollup, esbuild, tsc, babel, or a custom build command, Ovvoc runs it inside the isolated container.
The default build timeout is 60 seconds, configurable per repository in the dashboard settings. Build output is analyzed for errors versus warnings — warnings do not block the pipeline, but errors cause immediate abort and failure report generation.
Confidence Scoring
A score you can trust
100%
Version-Only Updates
No code changes needed. Version bump in package.json and lockfile regeneration only. Structurally identical to your current code.
95–100%
Deterministic AST Transforms
Code rewritten using AST-level transforms. Structurally verified, predictable output. High confidence because transforms are rule-based.
60–90%
AI-Assisted Transforms
Depends on transform complexity and your test coverage. Higher test coverage yields higher confidence. Human review recommended.
The confidence score is displayed in both the PR description and the dashboard. Use it to prioritize which PRs to review first — 100% confidence PRs can often be merged with minimal review, while lower-confidence PRs deserve more careful attention.
Ready to automate your dependency updates?
Start with one repo. See the difference in your first PR.