Projected from honua-sdk-js's sample catalog. Code is not vendored here -- follow the GitHub link below for the source.

Safe Agent Workbench

Proves typed proposals, shared policy validation, signed single-use approval, bounded execution, refusal, and verified receipt states.

Support tier: experimental Track: lab Protocols: geoservices, honua Renderers: none Lifecycle: active — North-star safe-agent flagship rebuilt around plan-bound approval and receipts.

Capabilities

Agent Operations (Validation Layer) (ai.agent-operations)Agent Approval Workflows (ai.approval-workflows)MCP Discovery & Query (ai.mcp-discovery)

View source on GitHub ↗ · Docs ↗