Reference
Key terms used in the documentation.
| Term | Meaning |
|---|---|
| Advanced intelligence | A setting that, when on, automatically routes a request to a more advanced model when needed. |
| Agent | An AI that plans toward a goal, calls tools, and evaluates the result. |
| Agentic | A way of working in which AI takes steps on its own and uses tools to reach a goal. |
| Apex | Desktop The service tier for the hardest planning and multi-step work; on Plus and Max. |
| Approval mode | Desktop The setting that determines which operations the agent asks approval for. |
| Auto | The tier where MORATUS picks the right service tier for each step; the default on Plus and Max. |
| Brain / worker | Desktop The division of labor between a strong model that plans and a fast model that does the reading. |
| Checkpoint | Desktop A record that lets you undo file changes. |
| Connector | A component that connects MORATUS to apps such as Gmail, Notion, and GitHub, or to creative tools. |
| Context window | The amount of text the model can consider at once (measured in tokens); 32K – 1M depending on plan. |
| Credit | The unit of usage in MORATUS. It is deducted based on actual usage and renews every period. |
| Default working mode | The setting that sets the reasoning level new chats start with. |
| Dictation | Turning your speech into text in the message box; you can edit it before sending. |
| Expert | The service tier for complex analysis, coding, and long-running work; on Plus and Max. |
| Fallback provider | Desktop The next provider in the chain, which a step moves to when a provider can't respond. |
| Fast | The service tier for everyday questions, writing, and summarizing; on every plan. |
| Guest mode | Web Limited use without an account; chats aren't saved. |
| Incognito chat | A chat that isn't saved to history, memory, or projects; used while signed in. |
| Library | The place where every file uploaded or generated in chats is collected. |
| Live voice chat | A real-time, two-way spoken conversation with MORATUS; its transcript is saved as a chat. |
| Memory | Lasting, useful information about you that MORATUS extracts from chats. |
| Merchant of Record | The party that collects payment, issues invoices, calculates taxes, and handles refunds; for MORATUS, Paddle. |
| OAuth | A standard sign-in method that grants access to an account or app without sharing a password. |
| Personalization | Preferences that set the tone of replies, how you're addressed, and your professional context, applied in every chat. |
| PKCE | A security method used in OAuth sign-ins to protect the authorization code from being stolen. |
| Plan mode | Desktop A mode in which the agent investigates and plans without changing anything. |
| Project | A workspace that brings chats, files, and instructions together. |
| Project notes | Notes MORATUS keeps about decisions made and progress in a project. |
| Prompt injection | An attempt to steer an AI agent with instructions hidden in a web page, document, or email. |
| Reasoning level | How much the model thinks before answering. A higher level gives more thorough replies; time and credit use go up. |
| Ref code | Desktop The code shown at the end of a reply that ended with an error, used to find the error's record. |
| RLS | Row-level security: ensures each user can access only their own records. |
| Sandbox | A secure environment where code runs isolated from the main system. |
| Scheduled task | A task that runs on the server by itself at the time you set. |
| Seat | The Lite, Plus, or Max plan assigned to a member on a Team plan. |
| Service tier | What you choose instead of a model: Fast, Expert, Auto, and Apex on Desktop. |
| Side chat | A separate chat that opens next to the main chat for a quick question without interrupting it. |
| Signed link | An address that opens only one specific file, and only for a short time. |
| Skill | A reusable package of instructions about a role, a method, and an output format. |
| Team plan | A plan several people use on one subscription and one invoice; starts at 3 seats. |
| Token | A small piece of text processed by the model; roughly part of a word. |
| Tool | A function the model can call: web search, file generation, reading a connector, and more. |
| Trust gate | Desktop The approval you give before the agent can work in a folder; the agent doesn't work in a folder you haven't approved. |