Taming The Claw: A Demo on Governing OpenClaw with Agent Control

OpenClaw is one of the most capable agent frameworks available today, and one of the easiest to lose control of.

Whether it’s a context window compaction that drops a safety instruction mid-run, or an agent suggesting that it be granted write access to your admin folder, this is what happens when safety is handled within the agent's context rather than enforced externally.

In this hands-on demo, our engineer, Lev Neiman, will walk through how to layer Agent Control on top of an OpenClaw deployment to close the governance gaps that prompt-based safety can't cover.

You’ll learn how to:

1. Install the Agent Control OpenClaw plugin

2. Set up Agent Control

3. Control tool calling with Agent Control in OpenClaw

What you'll leave with:

• A working Agent Control + OpenClaw integration you can adapt for your own stack

• Policy patterns for the most common OpenClaw failure modes: unconstrained tool access, permission escalation, uncontrolled sub-agents, and memory leakage

• A centralized control plane that your entire team — technical and non-technical — can update in minutes

Who this is for: Engineers actively building with or evaluating OpenClaw who want a practical path to production-grade governance.

Prerequisites: OpenClaw already installed and basic Python knowledge. A laptop with Docker installed. Setup instructions will be sent ahead of the session.