Skip to main content

​
The control plane for AI agents

Ship autonomous agents, without losing control. ActGuard adds runtime governance to agent systems. It sits between your agents and the APIs or tools they call, enforcing budgets, guardrails, and execution policies in real time. You integrate it directly in your agent code and manage policies from a central gateway. Framework-agnostic. Model-agnostic. Vendor-neutral.

​
What you can do with ActGuard

  • Enforce token and spending budgets per workspace, agent or plan
  • Add guardrails around tool usage
  • Send alerts to Slack or webhooks when policies are triggered
  • Inspect blocked runs and intervention events

​
Dashboard overview

Monitor agent runs and tool calls in real time
  • See token usage and spending across agents and workspaces
  • Inspect blocked executions when a policy is triggered
  • Configure budgets, guardrails, and alert channels
  • Review intervention events such as budget limits or policy violations
Instead of discovering problems after costs spike or tools are abused, ActGuard surfaces them while the agent is executing. Dashboard

Agent Management

Configure budgets and alerts, manage workspace keys, and inspect spend and guard interventions.

Python SDK

Start with budget_guard, then layer tool decorators. TypeScript support is coming soon.