chap
by BrightbeamAI
A MCP/A2A runtime for auditable human–agent handoffs and approvals
Summary
Implements a Collaborative Human Agent Protocol (CHAP) runtime for auditable human-in-the-loop workflows. Provides A2A Protocol Pattern compatible primitives for approvals, overrides, handoffs, escalation and verifiable evidence logs so every human–agent interaction is recorded and reproducible. Distinctive features include verifiable audit trails and explicit handoff/approval patterns to make human decisions first-class in multi-agent flows. Also leverages Model Context Protocol (MCP) compatible primitives to enhance context in human-in-the-loop interactions.
The Value Proposition
Ideal For
Teams building multi-agent systems that require verifiable human approvals, handoffs, and audit trails for trust, compliance, or post-hoc evaluation. Consider adopting practices from Role-Based Agent Pattern to standardize responsibilities and handoffs within your CHAP-enabled workflows.
Use Cases
- Capture and verify human approvals and overrides in multi-agent workflows
- Record tamper-evident evidence logs for agent-to-agent handoffs and escalations
- Provide audit trails to feed into reputation and continuous evaluation pipelines