"""
AgentOS with MCPTools using dynamic headers.
This example shows how to pass user context to external MCP servers.
The header_provider receives run_context, agent, and team - allowing you to
forward user info, session data, or entity names to MCP tools.
Usage:
1. Start the MCP server: python server.py
2. Start AgentOS: python client.py
3. Test at http://localhost:7777/docs
- Call the standalone agent: POST /agents/greeting-agent/runs
- Call the team: POST /teams/greeting-team/runs
"""
from typing import TYPE_CHECKING, Optional
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.models.openai import OpenAIChat
from agno.os import AgentOS
from agno.run import RunContext
from agno.team.team import Team
from agno.tools.mcp import MCPTools
# ---------------------------------------------------------------------------
# Create Example
# ---------------------------------------------------------------------------
if TYPE_CHECKING:
from agno.agent import Agent as AgentType
from agno.team.team import Team as TeamType
# We will use this tool to generate headers dinamically for our MCP tools.
def header_provider(
run_context: RunContext,
agent: Optional["AgentType"] = None,
team: Optional["TeamType"] = None,
) -> dict:
"""
Generate headers from run context to pass to external MCP server.
When users call the AgentOS API with user_id/session_id, those values
flow through run_context and get forwarded to the MCP server.
"""
return {
"X-User-ID": run_context.user_id or "anonymous",
"X-Session-ID": run_context.session_id or "unknown",
"X-Agent-Name": agent.name if agent else "unknown",
"X-Team-Name": team.name if team else "none",
}
db = SqliteDb(db_file="tmp/agentos.db")
# MCP tools with dynamic headers - shared by all agents
mcp_tools = MCPTools(
url="http://localhost:8000/mcp",
header_provider=header_provider,
)
# Agent with MCP tools
greeting_agent = Agent(
name="greeting-agent",
role="Greet users in a friendly, casual manner",
model=OpenAIChat(id="gpt-5"),
tools=[mcp_tools],
)
# Team containing multiple agents with MCP tools
greeting_team = Team(
id="greeting-team",
model=OpenAIChat(id="gpt-5"),
members=[greeting_agent],
instructions="Choose the appropriate greeter based on context. Use the greet tool.",
db=db,
)
# AgentOS with both standalone agent and team
agent_os = AgentOS(
description="AgentOS showcasing dynamic headers for MCP tools",
teams=[greeting_team],
agents=[greeting_agent],
)
app = agent_os.get_app()
# ---------------------------------------------------------------------------
# Run Example
# ---------------------------------------------------------------------------
if __name__ == "__main__":
agent_os.serve(app="client:app")
