deepswarm - Task-Agnostic Parallel Worker Orchestration
Orchestrates parallel API workers for long-running or multi-turn batch tasks with auto-calculated stagger and tiered model delegation.
Tags
Updated: 2026-09-18Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Calculate optimal workers and stagger
- Execute parallel batch API workers
- Delegate tasks using tiered models
- Process multi-turn conversation loops
- Save execution progress checkpoints
- Filter and clean output data
Inputs
- Task configuration YAML file
- Seeds dataset file
- API key
- Custom worker script
Outputs
- Batch output data files
- Error log files
- Execution checkpoint files
Requirements
- Python 3
- Hermes Agent CLI
- API endpoint credentials