CrewAI
Govern multi-agent CrewAI workflows with Nyraxis.
CrewAI Integration
Setup
import nyraxis_sdk
from crewai import Agent, Task, Crew
nyraxis_sdk.init(
api_key="nyx_...",
enforce=True,
instrument=True,
agent_name="crewai-research-crew",
)Example crew
researcher = Agent(
role="Research Analyst",
goal="Find accurate information about the given topic",
backstory="You are an expert researcher.",
llm="gpt-4o",
)
writer = Agent(
role="Content Writer",
goal="Write a clear summary based on research",
backstory="You are a skilled technical writer.",
llm="gpt-4o",
)
research_task = Task(
description="Research {topic}",
agent=researcher,
expected_output="Detailed research notes",
)
write_task = Task(
description="Write a summary based on the research",
agent=writer,
expected_output="A 500-word summary",
)
crew = Crew(
agents=[researcher, writer],
tasks=[research_task, write_task],
)
# Every LLM call by every agent is evaluated by Nyraxis
result = crew.kickoff(inputs={"topic": "AI safety regulations 2025"})What gets protected
- All LLM calls from all agents in the crew
- Tool calls and their outputs
- Inter-agent communication
- Final crew output
Per-agent policies
Assign different policies to different agents using agent names:
# In the Nyraxis dashboard, create policies scoped to specific agents
# e.g., "Block PII" only for the customer-facing agent