Terms of service

Last updated 2026-05-25 · sandbox version

1. Sandbox status

The service at sybilshield.org is a public sandbox. Scores are produced by a baseline model not yet calibrated on a full production corpus. Do not use sandbox output as a final filter for token distributions without manual review.

2. Acceptable use

• · Submit only public on-chain addresses
• · Don't use the service to dox individuals or violate any privacy law
• · Stay within the public sandbox fair-use limits (100 calls/month). Repeated overshoot may result in API key suspension
• · Don't share your API key or use it for >1 production system

3. No warranty

The sandbox is provided "as is" without warranty of any kind, express or implied. Scores carry inherent false-positive/false-negative rates documented on /methodology.

4. Liability

To the maximum extent permitted by law, SybilShield is not liable for any direct, indirect, incidental, or consequential damages arising from use of the sandbox, including but not limited to revenue loss, governance outcomes, or distribution decisions made based on scores returned by the service.

5. Appeals + audit-log

Any address holder can file an appeal at /appeal. We commit to respond within 48 hours. Audit-log records of score changes are append-only and reproducible from public methodology.

6. Open source

All six detection methods, the audit-log schema, and the appeal protocol are MIT-licensed on GitHub. You may fork, self-host, or vendor.

7. Termination

You may delete your account at any time by emailing support@sybilshield.org. We may suspend accounts that violate §2, with notice unless suspension is required to stop active abuse.

8. Changes

We'll publish material changes to these terms on this page and in /changelog at least 14 days before they take effect.

9. Governing law

Pre-incorporation: no specific jurisdiction is asserted. Post-incorporation, this will be updated to reflect the entity's registration jurisdiction.

Contact: support@sybilshield.org