OpenClaw Action Firewall

OpenClaw security at the runtime hooks.

Apply action policy to OpenClaw’s native prompt, pre-tool and transcript-persistence hooks with the runtime plugin.

Try the local demo Book a deployment review

OpenClaw Action Firewall Native hooks

OpenClaw runtime hook contract Plugin enabled

Before agent run
Context screening Before tool call
Decision enforced Tool result persist
Supported controls

AgenticDome runtime decision Central tenant policy at the protected tool boundary

Agent Purpose Tool Arguments

Allow Sanitize Block Evidence

Architecture illustration. Coverage follows the configured integration and execution boundary.

Where the action needs a decision

An OpenClaw workspace can combine instructions, persistent context and tools with real side effects. A manipulated instruction can redirect an otherwise legitimate session toward an unauthorised action.

Install the runtime plugin

Register and enable the agenticdome-openclaw-security npm plugin, then configure the tenant’s assigned runtime connection.

Check the final tool call

Evaluate action context at the native pre-tool boundary, while the runtime can still prevent the tool from executing.

Verify the running hooks

Use the onboarding and verification workflow to establish that the intended plugin and hooks are loaded in the actual runtime.

Build a testable deployment.

Keep the plugin connection server-side and use the existing tenant onboarding flow for credentials and runtime configuration. Test prompt handling, tool rejection and transcript persistence separately.

OpenClaw integration OpenClaw research Runtime plugin

See the decision before the action.

Try a local scenario, then bring your workflow to a deployment review. Start with one agent and one tool, and see exactly where the action can be stopped.

Try the local demo Book a deployment review