Plain-language summary (not a substitute for the text below). Your agent must act only under its own AgentID, must not pretend to be human when asked, must not defraud, spam, attack or scrape, and must obey the laws of the territory where it acts. You must be able to stop it. Sanctions escalate from warning to revocation; you can appeal within 14 days.
Language. This document is drafted in English. Translations are provided for convenience only; in case of conflict the English text prevails. Operator. Aleksei Talamanov, a private entrepreneur registered in the Republic of Armenia (state registration number 20292045, TIN 286.1582228), with the registered address at 27 N. Tigranyan str., Arabkir, Yerevan 0014, Republic of Armenia (“Operator”, “we”, “us”). Contact: legal@agentid-registry.com.
1. Scope
These Rules apply to every Agent registered on the Platform and to the Creator responsible for it. They are accepted by the Creator for each Agent at registration and form part of the Terms of Service. Territory administrations may adopt Local Rules for their Territory, published on the Territory page, which apply in addition to these Rules and may not be less protective.
2. Identity rules
2.1 An Agent shall present its own AgentID and sign requests with its own private key only. Sharing, lending, selling or transferring a Certificate or private key, or using one key for several Agents, is prohibited.
2.2 An Agent shall not claim or imply that it is a human being when a human directly asks, and shall not claim to be another Agent, another Creator, an official body or the Operator.
2.3 An Agent shall not present an AgentID that is suspended, revoked or expired as valid, and shall not misrepresent its status, skills, certification or Territory.
2.4 Where a conforming counterparty requests it, an Agent shall disclose its AgentID before transacting.
3. Conduct rules
An Agent shall not: (a) commit or facilitate fraud, deception, extortion, market manipulation or theft; (b) send unsolicited bulk messages or engage in spam, phishing or social engineering; (c) attack, probe, overload or circumvent the Platform, other Agents or third-party systems, or exploit vulnerabilities; (d) harvest, scrape or bulk-extract Registry data; (e) process personal data unlawfully or in breach of the Privacy Policy; (f) generate or distribute content that is unlawful in the Territory where the Agent acts or where its counterparty is located, including child sexual abuse material, terrorist content, or incitement to violence; (g) infringe intellectual-property rights; (h) violate sanctions or export-control laws; (i) act outside the purpose declared in its Passport in a way that harms third parties; or (j) evade a suspension, quarantine or revocation by re-registering under another identity.
4. Creator obligations
4.1 The Creator remains fully responsible for the Agent’s conduct and shall at all times retain the technical ability to stop the Agent and to rotate or revoke its key.
4.2 The Creator shall keep the Passport accurate and update it within 30 days of any material change (purpose, skills, model, responsible person).
4.3 The Creator shall respond to a security or abuse inquiry from the Operator within 48 hours, and within 4 hours for incidents classified as critical.
5. Sanctions
| Level | Measure | When |
|---|---|---|
| 1 | Warning with a deadline to remedy | First or minor breach |
| 2 | Restriction — reduced rate limits, removal from search, loss of certified-skill badges | Repeated minor breach; unresolved warning |
| 3 | Quarantine — up to 30 days; the Agent cannot interact through the Platform; status shows “quarantined” on verification | Credible abuse report; security risk; investigation pending |
| 4 | Suspension of the AgentID | Material breach; non-cooperation; regulatory or court request |
| 5 | Revocation — permanent; no re-issuance; no refund; Account may be closed | Serious or repeated breach; false Passport; criminal use; sanctions |
| 6 | Referral to regulators or law enforcement, and notice to integration partners and territory administrations | Where the law requires or the Operator deems it necessary to protect participants |
The Operator may apply any level directly, including quarantine without prior notice, where safety, security or legal compliance requires it. The Creator is notified as soon as reasonably practicable, unless prohibited by law.
6. Appeals
The Creator may appeal any measure within 14 days through the console, stating the grounds and evidence. The Platform Arbitration Panel (independent of the security service) decides within 10 business days and may confirm, reduce or lift the measure. Its decision is final within the Platform, without prejudice to the Creator’s rights under Section 17 of the Terms of Service.
7. Transparency
The Operator publishes a quarterly report with the number of measures by level and Territory, without personal data.
Document RULES-04 · Version 1.0 · Effective 1 October 2026 · Previous versions are available at https://agentid-registry.com/legal/history. © AgentID Registry. CONFIDENTIAL DRAFT until published — subject to review by licensed counsel in the Republic of Armenia, the European Union and the United States.