prose

Multi-agent orchestration coordinates concurrent AI sessions and block invocations using the OpenProse VM. You define agent templates and compose pipelines to execute complex workflows without bespoke glue code.

OpenProse VM is a semantic execution engine that manages multi-agent workflows. It handles state persistence, imports, and parallel constructs so teams can build complex AI pipelines with resumption and auditing capabilities.

To build multi-agent pipelines, create a SKILL.md file at your project root with a name and description. The OpenProse VM then executes your defined agents, blocks, and concurrent sessions across the workflow.

Yes, orchestrated AI sessions require a SKILL.md file at the project root containing the name and description. Optional scripts, references, and assets can be included alongside the OpenProse VM state backends.

State persistence and inspection are supported through integration with common OpenProse state backends. This enables workflow resumption, auditing, and debugging across concurrent sessions and map/reduce data flows.

Defining a single orchestrated workflow using OpenProse VM is the best approach. It coordinates data prep, model evaluation, and decision reasoning across multiple agents without requiring custom integration code.