DEV Community

João André Gomes Marques
João André Gomes Marques

Posted on

How to Add Governance to Your CrewAI Agents

CrewAI makes it easy to build multi-agent teams. But when those agents run in production, you need to know what they did and control what they can do.

The Problem

A CrewAI crew with 5 agents can make dozens of tool calls per run. Without governance:

  • No record of which agent did what
  • Any agent can call any tool
  • No way to prove compliance to auditors

Adding Asqav

pip install asqav
Enter fullscreen mode Exit fullscreen mode
import asqav

asqav.init(api_key="sk_...")

# Create identities for each agent in your crew
researcher = asqav.Agent.create("researcher")
writer = asqav.Agent.create("writer")
reviewer = asqav.Agent.create("reviewer")
Enter fullscreen mode Exit fullscreen mode

Signing crew actions

@asqav.secure
def search_web(query: str):
    return search_tool.run(query)

@asqav.secure
def write_report(data: dict):
    return writer_tool.run(data)
Enter fullscreen mode Exit fullscreen mode

Every tool call now gets a quantum-safe signature.

Policies for crews

# Only the researcher can search the web
asqav.create_policy(
    name="researcher-only-search",
    action_pattern="tool:search_web",
    action="block_and_alert",
    severity="medium"
)
Enter fullscreen mode Exit fullscreen mode

Links

Top comments (0)