Agent-agnostic by design

Govern any AI agent.

GOVENANT doesn’t govern “Claude agents” or “OpenAI agents.” It governs the evidence trail of autonomous work — so any agent whose actions hit an observable substrate can be measured, whatever framework, model, or vendor built it. The agent never has to “support GOVENANT.”

What runs the audit vs. what gets audited

The most common misread: the eight AI tools on the audit page are how you run the audit — not the limit of what it can measure. Keep the two separate.

What runs the audit

ClaudeChatGPTClaude CodeCursorCodexVS CodeWindsurfCline

An AI client you already use reads the open instrument and drives the read-only sweep. This is the runner — not the thing being judged.

What gets audited

AgentforceLangGraphRetelln8nCrewAIa custom Python agenta browser/computer-use agenta robotics stack

Any agent system with an observable record of actions and outcomes — whatever built it. The agent never has to "support GOVENANT."

Ground truth: the database + actions + outcomes + duties + gates — the substrate, never the logs’ self-report.

By agent class

How the agents are built — and the natural substrate each class already produces.

By use case

What the agents do — the industry and function where an unverified “done” costs the most.

Don’t see your agent? It probably still works.

If your agent’s actions land in a record you can read, GOVENANT can measure it. Run the free audit from the tool you already use, or read how the standard governs substrates rather than frameworks.