PatchBoard: Schema-Grounded State Mutation for Reliable and Auditable LLM Multi-Agent Collaboration
For developers of LLM multi-agent systems, PatchBoard provides a reliable and auditable alternative to natural-language coordination, significantly improving task success and token efficiency.
PatchBoard introduces a schema-grounded collaboration architecture for LLM multi-agent systems that replaces inter-agent dialogue with validated JSON Patch mutations over a shared structured state, achieving an 84.6% success rate on ALFWorld (vs. 30.8% for LangGraph and 61.6% for Flock) while reducing tokens per successful task to 45.5k (vs. 368.3k and 64.2k).
LLM multi-agent systems often coordinate through natural-language dialogue or loosely structured shared memory, making intermediate state difficult to validate, attribute, and audit. We introduce PatchBoard, a schema-grounded collaboration architecture that replaces inter-agent dialogue with validated JSON Patch mutations over a shared structured state. An Architect agent constructs a task-specific schema and workflow rules, while a deterministic kernel validates each proposed state mutation against schema constraints, role-specific write contracts, and runtime invariants before committing it transactionally. On 630 matched ALFWorld episodes, PatchBoard achieves an 84.6% success rate, compared with 30.8% for LangGraph and 61.6% for Flock, while reducing tokens per successful task to 45.5k, compared with 368.3k and 64.2k, respectively.