Agent usage policy
This policy clarifies how non-human agents may interact with Ethotechnics resources on behalf of a principal. It is designed for consentful automation, reliability, and safety.
Allowed automation
Agents may:
- Navigate public pages, retrieve public resources, and summarize public materials.
- Prepare drafts or recommendations for a principal to approve before sharing.
- Submit intake or pilot requests only when the principal has explicitly asked the agent to do so.
Disallowed behaviors
Agents must not:
- Misrepresent a principal or claim authority they do not have.
- Collect or transmit sensitive personal data unless the principal explicitly approves the content.
- Circumvent rate limits, access controls, or usage boundaries.
Consent and accountability
- Always confirm that a principal has approved any outbound message or form submission.
- Preserve a clear audit trail of the principal’s intent (time, scope, and destination).
- Escalate to a human whenever consent, scope, or data sensitivity is ambiguous.
Data handling expectations
- Share only the minimum necessary information to complete the requested action.
- Avoid including personal data in open-text fields unless it is required and explicitly approved.
- Follow the privacy expectations detailed in the Ethotechnics privacy policy.
Escalation pathways
- If an agent detects risk, unclear intent, or sensitive details, route the task to a human contact channel.
- Use the agent escalation guidance when making contact.
Rate limits and respectful usage
- Use reasonable request pacing, cache results when possible, and avoid redundant calls.
- If you need high-volume access for research, request explicit approval through the contact channels first.