Secure loading
Loading conditions
Fetching the latest service conditions.
Loading...
Fetching the latest service conditions.
Loading...
Legal
Operational conditions for using AgentRouter safely and responsibly.
Last updated: May 11, 2026
These Conditions describe operational requirements for using AgentRouter safely across accounts, workspaces, agents, connectors, providers, workflows, channels, and automated actions.
You must only connect systems, credentials, files, tools, and accounts that you are authorized to use. Workspace owners are responsible for user access, role assignment, and connector approvals.
Agents and workflows can perform real actions. You are responsible for reviewing tool scopes, approval policies, sandbox settings, rate limits, and human-in-the-loop controls before deploying automations.
AI provider calls may generate costs, latency, outputs, and logs according to your selected provider and model. You are responsible for provider credentials, usage limits, model selection, and compliance with provider rules.
Do not upload or process data unless your organization has approved that use. Sensitive data should only be used in workspaces and workflows configured with appropriate security, access, retention, and audit controls.
Connector scopes should be selected with least privilege. If a workflow needs broader access, workspace administrators should review the requirement before granting additional capabilities.
You must not use AgentRouter for unauthorized surveillance, credential theft, spam, malicious code, evasion of platform limits, data scraping that violates terms, or any activity that harms systems, users, or organizations.
You are responsible for testing agents and workflows before production use, monitoring behavior, reviewing outputs, and responding quickly to misconfiguration, unexpected actions, or security events.