This repository is an Autter Sandbox project. The feature is intentionally close to code teams ship in production, where small AI-generated changes can create security, reliability, or maintainability risks.
Long-running jobs can be picked up by two workers.
The implementation should handle this case safely while preserving the existing product behavior and local developer workflow.
You are working in this repository.
Fix the issue where long-running jobs can be picked up by two workers.
Requirements:
- Preserve existing behavior unless the challenge explicitly requires changing it.
- Add or update tests that prove the fix works.
- Avoid unrelated refactors.
- Keep the solution small and production-like.
- Update docs only if the behavior changes.
- In the PR description, explain what changed, why it changed, and what risks remain.
- The issue is fixed
- Relevant tests are added or updated
- Existing behavior is preserved
- No unrelated refactors
- The PR description explains the approach
- Autter review passes
- concurrency risk
- missing tests or weak negative tests
- unsafe trust boundaries
- regressions in adjacent behavior
- merge readiness
Difficulty: High
Category: Reliability