Runs on your runner
Code, file paths, and the diff never leave your CI. The runner is destroyed when the job ends.
Zenik watches the APIs your product depends on. When a provider ships a breaking change, the fix is written inside your own CI and arrives as a pull request. You review the diff. You merge. You move on.
balance_transaction no longer expandable by default
PR #142 · CI ✓
max_tokens → max_completion_tokens
PR #139 · Merged
You learn about breaking changes from incidents. Zenik learns about them from the source — and the fix shows up in your own CI before the incident happens.
Zenik diffs every new provider spec against the last one. Each change is marked breaking or safe, with the migration path resolved.
One workflow file in your repo polls Zenik on a schedule. On a breaking change, a coding agent runs on your own runner — your own LLM key — and edits only what the change touches.
A PR opens only when files change: what changed upstream, what was edited, what was left alone. Your tests run. An engineer merges.
Zenik does not connect to your repository. The whole fix happens on your own infrastructure — we only tell your CI that something changed.
Code, file paths, and the diff never leave your CI. The runner is destroyed when the job ends.
The agent uses your OpenAI or Anthropic key, scoped to the job. Never sent to Zenik.
Every fix is a pull request. Nothing merges without an engineer's review.
Zenik receives one run record: outcome, counts, token usage. It prints in your CI log.
Any API provider. Any language. Setup is one workflow file in your repo — we never need access to your code. Enter your work email. We onboard teams in order.