Praxa Harness Formalizes AI Agent Execution with Evidence Trails

VERA Brief

AI-generated. Grounded in the article and its cited sources.

Researchers have introduced Praxa, an agent harness that provides explicit representation and verification for AI agent actions, from proposal to external effect. The system uses evidence lanes to track execution states and has undergone initial testing.

Key facts

• Praxa is an agent harness designed to represent and verify AI agent execution states.
• The system manages states through deterministic admission, brokered execution, external read-back, reconciliation, and reviewed promotion.
• Four evidence lanes were reported as part of the system.
• A pilot of Praxa across 12 tasks showed mixed results in strict trials.
• A source-authored candidate agent demonstrated comparable accuracy with fewer tokens and lower estimated costs.