CrewAI Integration

Route all CrewAI agent LLM calls through the Curate-Me gateway for cost tracking, rate limiting, and governance. CrewAI uses LiteLLM under the hood, so you can configure the gateway via environment variables or per-agent config.

Before and after

Environment variable approach (easiest)


# Before — direct to OpenAI
export OPENAI_API_KEY=sk-your-openai-key
 
# After — through Curate-Me gateway
export OPENAI_API_BASE=https://api.curate-me.ai/v1/openai   # ← added
export OPENAI_API_KEY=sk-your-openai-key

Your existing CrewAI code needs zero changes:


from crewai import Agent, Task, Crew
 
researcher = Agent(
    role="Research Analyst",
    goal="Find key facts about AI governance",
    backstory="You are a senior analyst.",
    llm="gpt-4o",
)
 
task = Task(
    description="List 3 reasons teams need an LLM gateway.",
    expected_output="A bullet-point list with explanations.",
    agent=researcher,
)
 
crew = Crew(agents=[researcher], tasks=[task])
result = crew.kickoff()  # ← governed by Curate-Me

Per-agent configuration (more control)

For per-agent cost tagging and explicit gateway auth, use llm_config:


from crewai import Agent, Task, Crew
 
researcher = Agent(
    role="Research Analyst",
    goal="Find key facts about AI governance",
    backstory="You are a senior analyst.",
    llm="gpt-4o",
    llm_config={
        "api_key": "sk-your-openai-key",
        "base_url": "https://api.curate-me.ai/v1/openai",
        "extra_headers": {
            "X-CM-API-Key": "cm_sk_YOUR_KEY",
            "X-CM-Tags": "project=crewai-app,team=research",
        },
    },
)
 
writer = Agent(
    role="Technical Writer",
    goal="Summarize the research findings",
    backstory="You are an experienced technical writer.",
    llm="gpt-4o",
    llm_config={
        "api_key": "sk-your-openai-key",
        "base_url": "https://api.curate-me.ai/v1/openai",
        "extra_headers": {
            "X-CM-API-Key": "cm_sk_YOUR_KEY",
            "X-CM-Tags": "project=crewai-app,team=writing",
        },
    },
)
 
research_task = Task(
    description="List 3 reasons teams need an LLM gateway.",
    expected_output="A bullet-point list with explanations.",
    agent=researcher,
)
 
writing_task = Task(
    description="Write a 2-paragraph summary of the research.",
    expected_output="A concise summary suitable for a blog post.",
    agent=writer,
)
 
crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
)
result = crew.kickoff()

With per-agent X-CM-Tags, you can see exactly how much each agent spent in the Cost Tracking dashboard.

Prerequisites


pip install crewai

What you get

Every CrewAI agent call through the gateway automatically receives:

Feature	Description
Cost tracking	Per-agent token and dollar cost via `X-CM-Tags`
Rate limiting	Per-org, per-key request throttling
PII scanning	Regex scan for secrets and PII before hitting the provider
Model allowlists	Only approved models per your org policy
Budget caps	Daily and monthly spend limits enforced
Audit trail	Full request metadata logged for compliance

Per-agent cost attribution

Use X-CM-Tags to break down costs by agent, team, or project:


X-CM-Tags: project=crewai-app,agent=researcher,team=data
X-CM-Tags: project=crewai-app,agent=writer,team=content

View tagged costs in the dashboard under Gateway > Cost Tracking > Breakdown.

Working example

Python example — multi-agent crew with per-agent cost tags

Next steps

Gateway Quickstart — full setup walkthrough
LangChain Integration — LangChain chains and agents
Cost Tracking — budget alerts and cost attribution