Demo Environment
The MCP Runtime
Arcade delivers secure agent authorization, high-accuracy tools, and centralized governance for multi-user AI agents at scale.
Agent Authorization
User-specific permissions that integrate with your OAuth and IDP flows
Agent-Optimized Tools
1000s of MCP tools built for agents, not just API wrappers
Centralized Control
One place to manage access, versioning, and audit logs across all tools
from arcadepy import Arcade
client = Arcade()
user_id = "user@example.com"
# Start authorization for this user
auth = client.auth.start(
user_id, "google",
scopes=["gmail.readonly"]
)
# Agent can now act on behalf of user
response = client.tools.execute(
tool_name="Google.ListEmails",
inputs={"n_emails": 5},
user_id=user_id
)
# ✓ Returns user's emails securely