Step Actions Reference

To configure AI agent workflow steps for deterministic execution, define specific action types and parameters such as `agent.run` for single-agent tasks, apply global settings like `priority`, and set limits like `stepMaxDurationMinutes`.

In automated workflows, `agent.run` handles single-agent task execution, while `agent.session` coordinates multi-agent efforts. Selecting the correct action type ensures your AI agents perform tasks with the right scope and parameters.

To use `git.clone` and `repo.identify` in a CI/CD workflow, configure the specific `cloneOptions` and repository parameters before executing subsequent steps like `ci.executeWorkflow` for continuous integration.

Yes, you can set execution timeouts and priority limits for AI agent tasks using workflow-level settings like `priority` and `timeoutMs`, alongside step-specific constraints such as `stepMaxDurationMinutes`.

Limitations when configuring multi-agent sessions in CI pipelines include step duration constraints like `stepMaxDurationMinutes` and global timeout settings, which bound execution time and require precise parameter configuration to avoid failures.

Yes, Step Actions Reference provides detailed documentation on Overcut workflow step action types, parameters, and configuration options, covering agent execution, repository operations, and CI integration.