When a dependency or API upgrade breaks your build, the Sunup agent starts right away. It rewrites the affected code, verifies the fix with your own tests in an isolated sandbox, and opens a pull request.
Free, 20 minutes. We go through your open Dependabot and Renovate pull requests with you, then send a short report on what's failing and what Sunup could fix.
Every API sunset needs a sunup.
APIs keep changing. Sunup makes sure your code rises with them.
Animated example: the Sunup agent reads the OpenAI SDK migration guide, finds four outdated lines in app/chat.py, rewrites them, runs 212 tests in a sandbox, and opens pull request number 214.
EXAMPLE TRACEopenai-python 0.28 → 1.x·resolved in 3m 14s·verified in an isolated sandbox
Sunup keeps a live map of every package and SDK your code depends on, from the big providers to the smallest library. When one ships a breaking change, it already knows which lines are affected and goes to work.
Everything a careful engineer would do for a migration, done the same way every time.
Sunup tracks changelogs, SDK releases, and deprecation notices for the packages you depend on, and maps each breaking change to the exact lines in your code that use it.
Known changes are applied by deterministic syntax-tree rules. AI handles only what needs judgment, with small, focused context. Nothing else in your code moves.
Your own build, type checks, and tests run in a single-use container with networking switched off. If a changed line isn't covered by a test, the pull request says so.
Every fix arrives as a regular GitHub pull request with the upstream notice, the diff, and the test evidence. Review and merge it like any other PR.
The upgrades your team keeps postponing start the moment a breaking change is detected.
A new version ships with a breaking change. The upgrade pull request fails CI, and GitHub alerts Sunup instantly.
Sunup reproduces the failure in a sandbox and pinpoints every affected line using compiler errors and syntax-tree search.
The fix runs against your own test suite inside a secure, single-use container.
A passing pull request lands in GitHub with the evidence, turning a potential outage into a routine merge.
Sunup proposes changes. People decide what ships.
Every change arrives as a pull request and goes through your normal review.
Sunup never edits or deletes your tests to make them pass. If a test looks wrong, it tells you instead.
Each pull request shows which tests ran and which changed lines they covered.
If Sunup isn't confident a change is correct, it tells you instead of opening a pull request.
Sunup reads proprietary source code, so every step is designed to keep it contained. Here's exactly what happens to yours.
Install Sunup on the repositories you choose, not your whole organization. It asks GitHub only for the permissions it needs, and not for permission to change CI workflows, so GitHub itself blocks it from editing them.
Every run gets a fresh container. Dependencies install first, then networking is switched off before your build and tests run. No secrets, no production access, and the container is destroyed when the run ends.
Sunup works on a temporary copy of your code that's deleted after each run, and keeps only the diff and logs that explain a pull request, for 30 days. Code sent to our AI provider is covered by terms that don't allow training on it.
Need a security questionnaire answered or a walkthrough before you install? We'll go through it with your team. A formal SOC 2 audit is planned as we onboard paying customers.
Sunup moves your code off the version being retired and onto the one replacing it, then proves the change with your own tests.
Animation: old package versions set below a horizon line while their upgraded versions rise, for example openai 0.28 setting and openai 1.x rising.
Book a free upgrade audit. We'll go through your open upgrade pull requests with you and show what Sunup would fix first.
We'll reply within 2 business days to find a time.