MasterOS

Reference

Glossary

Plain-language definitions, work examples and references to the relevant modules.

  1. Prompt
    In this program
    A request or set of instructions you give an AI tool.

    Example: Ask AI to draft a reply using the supplied ticket and refund policy.

    M3 - Write instructions for AI

  2. Skill
    In this program
    A reusable way of working saved with a name, instructions, inputs and checks for the result.

    Example: Save a customer-reply Skill with the information it needs and the promises it must not make.

    M5 - Save and evaluate a reusable way of working

  3. Knowledge
    In this program
    Information sources AI can retrieve to help answer a question.

    Example: Use the applicable refund policy as a source, then check that it supports the answer.

    M2, M7 - Prepare information and use Knowledge

  4. Citation
    In this program
    A reference to a source used for an answer. It helps you check a claim; it does not guarantee that the claim is correct.

    Example: Open the cited document and compare the important claim with the relevant passage.

    M2, M7 - Check sources

  5. Model / model selection mode
    In this program
    A model generates the response. A selection mode guides how the system chooses a model; it is not a specific model name.

    Example: MasterOS offers Smart, Fast and Best. Record the underlying model only when it is shown.

    M3 - Choose a model for the task

  6. Quota
    In this program
    The usage allowance under your plan. Check your remaining allowance before starting work.

    Example: Review the remaining allowance and the cost of repeated runs when comparing ways of working.

    M0, M3 - Usage and cost

  7. Agentic AI
    In this program
    An approach in which AI agents pursue goals within controlled workflows, using Skills, sources and tools. Human decisions, supervision and recovery remain explicit; multiple agents are optional, not a higher level that every task requires.

    Example: An agent chooses the relevant lookup, a workflow checks the result, and a person reviews a consequential action before it is executed.

    M9-M10 - Agent decisions and controlled operation

  8. Agent
    In this program
    An AI system that works toward a goal by choosing steps and permitted tools within a defined scope. It needs access boundaries, stopping limits and an accountable owner.

    Example: An agent checks permitted order information and drafts a reply. Sending a message or changing a record still follows the required approval rules.

    M9-M10 - Choose, evaluate and operate an agent

  9. Workflow
    In this program
    A sequence of work steps with defined inputs, outputs, owners and conditions for moving forward. Skills, sources and tools can be connected, with explicit checks and exception paths.

    Example: A support request passes through a summary Skill, a reply-drafting Skill and a review step. Missing information stops the flow and goes to the responsible person.

    M6 - Connect Skills into a workflow