Add policy enforcement for cloud browser sessions

[L1AD]

This server gives agents control of cloud browser sessions via Stagehand:

• Session management -- starts cloud browser instances (billable)
• navigate -- visits arbitrary URLs
• click_element -- triggers UI actions on any page

Cloud browser sessions are billable. An agent loop starting sessions can generate unexpected costs. Navigation and click tools mean the agent can interact with any website including authenticated sessions.

Suggestion: Consider documenting support for PolicyLayer/Intercept, an open-source MCP proxy that enforces policies on tool calls before execution.

Example policy:

version: "1"
default: allow

tools:
  start:
    rules:
      - rate_limit: 5/hour

  click_element:
    rules:
      - rate_limit: 30/minute

  navigate:
    rules:
      - rate_limit: 20/minute

One line to set up: npx -y @policylayer/intercept init

[denial123789]

This is a strong example of why browser automation is becoming runtime infrastructure for agents, not just another tool call.

For cloud browser sessions, I would expect teams to need a few runtime-level controls around loops:

• policy declared before a session starts
• enforced denials during navigation/actions
• structured events for what was allowed, denied, or timed out
• replay/debug information after the run
• budget or rate limits for sessions started by an agent loop

We included Browserbase in Awesome Agent Runtime under Browser Automation, and this policy-enforcement angle is exactly why it belongs in the agent infrastructure map:
https://github.com/sandbaseai/awesome-agent-runtime

If the project has a preferred wording for this category, happy to adjust the entry.