What problem does it solve?
This Skill prevents Codex from proceeding until long-running external tasks, such as cluster jobs or batch pipelines, have successfully completed, ensuring downstream processes have the necessary prerequisites.
Core Features & Use Cases
- Asynchronous Task Monitoring: Polls the status of external jobs at a configurable interval.
- Flexible Completion Criteria: Supports completion based on command exit codes or output matching specific regular expressions.
- Robust Timeout and Retry Logic: Includes hard timeout caps and handles transient transport failures with retries.
- Use Case: You've submitted a large data processing job to a cluster. Use this Skill to pause your workflow until the job is done, then automatically proceed with analyzing the results.
Quick Start
Use the wait-for-job skill to poll the command 'kubectl get job my-job -n my-namespace -o jsonpath='{.status.conditions[*].type}'' for the 'Complete' status, checking every 120 seconds.