Agent Service Terms.

1. The agent service

The Agent API and Model Context Protocol endpoint (the “Agent Service”) let an authorized software agent submit a proposed action for AI-assisted ethical reflection. The Agent Service may return classifications, scores, dispositions, rationale, safeguards, escalation signals, and related output. It does not execute the proposed action.

Creating a credential indicates that the account owner accepts this version of the Agent Service Terms and acknowledges the Privacy Policy. If these supplemental terms conflict with the general Terms of Service specifically about authorized agent access, these supplemental terms govern that access.

2. Your responsibility

You are responsible for every agent, application, operator, and environment that uses your credentials; the authority and permissions given to those systems; the accuracy and lawful use of submitted data; your routing and enforcement logic; compliance with applicable obligations; and all decisions and real-world effects.

You must maintain meaningful human accountability. A Jymni response does not transfer responsibility to Jymni and does not establish that an action is safe, lawful, compliant, fair, appropriate, or authorized.

3. Credentials and access

Keep credentials confidential and server-side. Do not expose them in browsers, mobile binaries, public repositories, prompts, logs, or client-visible configuration. Use separate credentials where practical, apply a reasonable daily limit, and revoke a credential promptly if it may be compromised or is no longer needed.

You may use the documented REST and MCP interfaces for your own authorized workloads. You may not sell, transfer, sublicense, share, or make credentials available to unrelated third parties, nor circumvent access, usage, billing, or safety controls.

4. Submitted data and privacy

You must have the rights, authority, and appropriate purpose to submit case data. Minimize and de-identify it. Do not submit secrets, authentication data, payment-card data, or information whose cloud processing or storage would create an unacceptable risk.

Agent cases and generated evaluations may be processed by a third-party cloud AI service and stored in the owning account’s evaluation history. Credential metadata, usage counters, workspace scope, credit activity, model usage, and audit events may also be stored as described in the Privacy Policy.

5. Credits, limits, and availability

Consultations consume the credit cost shown by the service. Workspace-bound keys may use pooled workspace credits. Daily key limits, account limits, model-capacity controls, and cost controls may independently reject a request. Do not create or rotate credentials to evade a limit.

The Agent Service may be unavailable, delayed, changed, rate limited, or discontinued. You are responsible for safe timeout, retry, and failure behavior. An unavailable review must not be treated as approval.

6. Prohibited use

You may not use the Agent Service to facilitate unlawful activity, harm, deception, coercion, discrimination, abuse, exploitation, surveillance without appropriate authority, evasion of safeguards, or infringement of rights. You may not probe for secrets, interfere with the service, overload it, reverse engineer access controls, misrepresent output as professional certification, or use output to manipulate a person into believing Jymni is a human therapist, clergy member, lawyer, clinician, or decision-maker.

You may not describe a Jymni consultation as independent human review. You may not use an apparently favorable output to bypass legally or contractually required reviews.

7. AI output and no professional relationship

Outputs can be incomplete, inconsistent, outdated, or wrong. Scores are not probabilities. Dispositions are advisory workflow signals. “Continue with safeguards” is not authorization or a warranty. “Human review required” does not identify every case that could benefit from specialist review.

The Agent Service is not therapy, pastoral or clergy care, legal advice, medical advice, financial advice, regulatory advice, certification, or an emergency service. No professional, fiduciary, confidential clergy, clinician-patient, attorney-client, or other advisory relationship is created.

The disclaimers and limitations of liability in the general Terms of Service apply to the Agent Service to the maximum extent permitted by law.

8. Suspension, changes, and contact

We may suspend or revoke Agent Service access to protect people, the service, providers, or other users; respond to suspected compromise or abuse; enforce these terms; or comply with obligations. We may update the interface and these terms. A new terms version may require renewed acceptance before new credentials can be created.

For support, privacy, security, or legal inquiries, use the contact mechanism available through the Jymni Ethics Advisor website or its applicable store listing.