Liminalis
Legal

Acceptable Use Policy

Draft · last updated PLACEHOLDER: date

⚠
Draft. Section structure is in place; operative wording is marked with placeholders and has not been reviewed by a lawyer. This document is not yet in force.

1. Scope

This policy applies to all use of the Liminalis service, including the web application, the desktop client, and the agent and Model Context Protocol interfaces. It forms part of the Terms of Service.

2. Prohibited content and conduct

PLACEHOLDER: the prohibited categories. At minimum this needs to cover: unlawful material; child sexual abuse material; content that sexualises minors in any form, including fictional; targeted harassment; incitement to violence; material designed to deceive about its origin or authorship; and infringement of others' intellectual property.

Liminalis is used heavily for fiction, including dark fiction. The policy needs to draw its lines in a way that is workable for creative work while remaining unambiguous about the categories above. PLACEHOLDER: that boundary, stated explicitly.

3. Content you upload

You are responsible for having the rights to the material you place in a project. Do not upload material you are not permitted to process, including confidential material belonging to someone else and personal data you have no lawful basis to hold. PLACEHOLDER: any specific restrictions on regulated categories of personal data.

4. Corpora containing personal data

Some lenses are designed for public-record corpora that name real people — legislators, sponsors, officials. Where the service supports exporting such material, features that name individuals are off by default and must be enabled deliberately. PLACEHOLDER: obligations when redistributing exported material.

5. Technical restrictions

Do not attempt to circumvent metering, access controls, rate limits or the retrieval barriers that separate one user's material from another's. Do not use the service to attack, probe or overload any system, ours or anyone else's. PLACEHOLDER: rate limit specifics and automated-access rules.

6. Agents and automated use

If you connect an agent to the service through an API key, you remain responsible for what it does. Scope keys to the minimum needed; write access is off by default and should stay that way unless you have a reason.

7. Shared projects

Access control in shared projects is enforced before material is assembled into a model request, not filtered out of a response afterwards. Do not attempt to defeat it, and do not share material into a project in a way that misrepresents who may see it.

8. Reporting

To report misuse, see Support. PLACEHOLDER: dedicated abuse reporting address and expected response time.

9. Enforcement

PLACEHOLDER: the range of responses — warning, feature restriction, suspension, termination — how a decision is communicated, and how to appeal it.