Create data science agents

Designing parallel-agent workflows for data science decision-packs involves configuring structured YAML agent prompts that enable multi-strategy exploration across different tools and priors. This approach coordinates orchestrators and subagents to evaluate multiple forecasting strategies simultaneously.

A data science orchestrator coordinates subagent prompts to execute repeatable analytics workflows within decision-packs. It uses structured frontmatter with name and description fields to define mode behaviors and route tasks across parallel subagents for multi-strategy exploration.

Yes, you can configure a parallel-agent workflow to run multiple forecasting strategies simultaneously. By setting up an orchestrator that coordinates subagents evaluating different priors, the decision-pack ecosystem supports multi-strategy exploration and transparent failure handling across different prompts and tools.

Structuring agent prompts for transparent failure handling requires consistent frontmatter defining names and descriptions, alongside structured YAML for mode definitions. Documenting how to integrate tools, skills, and workflows within the decision-pack ecosystem ensures repeatable templates and transparent diagnostics across parallel-agent configurations.

No specific dependencies are required to set up multi-strategy exploration with decision-packs. You can create an orchestrator prompt at the agents directory using structured YAML frontmatter to define parallel-agent workflows, enabling multi-strategy exploration and transparent failure handling without external dependencies.