Advanced Scopes

import os
from datetime import UTC, datetime, timedelta

import jwt
from agno.agent import Agent
from agno.db.postgres import PostgresDb
from agno.models.openai import OpenAIResponses
from agno.os import AgentOS
from agno.os.config import AuthorizationConfig
from agno.tools.websearch import WebSearchTools

JWT_SECRET = os.getenv("JWT_VERIFICATION_KEY")
if not JWT_SECRET:
    raise RuntimeError("Set JWT_VERIFICATION_KEY before starting AgentOS")

db = PostgresDb(db_url="postgresql+psycopg://ai:ai@localhost:5532/ai")

# Create multiple agents with different capabilities
web_search_agent = Agent(
    id="web-search-agent",
    name="Web Search Agent",
    model=OpenAIResponses(id="gpt-5.2"),
    db=db,
    tools=[WebSearchTools()],
    markdown=True,
)

analyst_agent = Agent(
    id="analyst-agent",
    name="Data Analyst Agent",
    model=OpenAIResponses(id="gpt-5.2"),
    db=db,
    markdown=True,
)

admin_agent = Agent(
    id="admin-agent",
    name="Admin Agent",
    model=OpenAIResponses(id="gpt-5.2"),
    db=db,
    markdown=True,
)

# Create AgentOS with RBAC
agent_os = AgentOS(
    id="my-agent-os",
    name="RBAC Scope Demo",
    agents=[web_search_agent, analyst_agent, admin_agent],
    authorization=True,
    authorization_config=AuthorizationConfig(
        verification_keys=[JWT_SECRET],
        algorithm="HS256",
    ),
)

app = agent_os.get_app()


def create_token(user_id: str, scopes: list[str], hours: int = 24) -> str:
    """Helper function to create JWT tokens."""
    return jwt.encode(
        {
            "sub": user_id,
            "scopes": scopes,
            "exp": datetime.now(UTC) + timedelta(hours=hours),
            "iat": datetime.now(UTC),
        },
        JWT_SECRET,
        algorithm="HS256",
    )


if __name__ == "__main__":
    # 1. ADMIN - Full access to everything
    admin_token = create_token("admin_user", ["agent_os:admin"])

    # 2. POWER USER - Global access to all agents
    power_user_token = create_token(
        "power_user",
        ["config:read", "agents:read", "agents:run", "sessions:read", "sessions:write"],
    )

    # 3. LIMITED USER - Only specific agents
    limited_user_token = create_token(
        "limited_user",
        [
            "agents:web-search-agent:read",
            "agents:web-search-agent:run",
            "agents:analyst-agent:read",
            "agents:analyst-agent:run",
        ],
    )

    # 4. READ-ONLY USER - Can view but not run
    readonly_user_token = create_token(
        "readonly_user",
        ["agents:*:read", "config:read"],
    )

    # 5. WILDCARD USER - Can run any agent
    wildcard_user_token = create_token(
        "wildcard_user",
        ["agents:read", "agents:*:run"],
    )

    # These are local test tokens. Do not print production credentials.
    print("1. ADMIN (full access):", admin_token)
    print("2. POWER USER (global access):", power_user_token)
    print("3. LIMITED USER (specific agents):", limited_user_token)
    print("4. READ-ONLY USER (view only):", readonly_user_token)
    print("5. WILDCARD USER (run any):", wildcard_user_token)

    agent_os.serve(app="advanced_scopes:app", port=7777, reload=True)

uv pip install -U agno openai pyjwt ddgs "fastapi[standard]" uvicorn sqlalchemy pgvector "psycopg[binary]"

docker run -d \
  --name agno-postgres \
  -e POSTGRES_DB=ai \
  -e POSTGRES_USER=ai \
  -e POSTGRES_PASSWORD=ai \
  -p 5532:5432 \
  pgvector/pgvector:pg17

python advanced_scopes.py

# Limited user sees only 2 agents (not admin-agent)
export LIMITED_TOKEN="<limited_user_token>"
curl -H "Authorization: Bearer $LIMITED_TOKEN" http://localhost:7777/agents

# Read-only user cannot run agents (403 Forbidden)
export READONLY_TOKEN="<readonly_user_token>"
curl -X POST -H "Authorization: Bearer $READONLY_TOKEN" \
  -F "message=test" http://localhost:7777/agents/web-search-agent/runs

# Wildcard user can run any agent including admin-agent
export WILDCARD_TOKEN="<wildcard_user_token>"
curl -X POST -H "Authorization: Bearer $WILDCARD_TOKEN" \
  -F "message=Hello" http://localhost:7777/agents/admin-agent/runs