PersonaStack terms in plain language

These are the names you will see in the app. Each one describes a different part of how an agent does its work.

Accounts, workspaces, and organizations

Account: your sign-in identity and profile.
Workspace: the place where you keep personas, stacks, AI providers, and integrations. Select it in the sidebar before working.
Organization: the business account that manages its workspaces, members, access settings, and shared plan.
Administrator: a person with management access at a particular level. Workspace and organization roles are separate.

Persona: an AI agent with a name, instructions, and selected tools.
Personality and Instructions: guidance that describes how the persona should normally work.
Stack: a team of personas with a shared mission.
Mission: the outcome the team should deliver.
Responsibilities: the work assigned to one member of the stack.
Rules: instructions shared by the whole stack.
Stack Stream: the team’s shared conversation.
Stack template: a reusable starting point for creating a team with your own providers and connections.

AI provider: the saved connection used to access a model service or supported provider subscription.
Model: the AI selected on that provider connection.
AI Agent Location: where PersonaStack runs the agent. It does not change where a cloud provider runs its model.
Integration: a saved connection to a service or device.
Assignment: selecting that integration for a particular persona.
Capability: a built-in ability enabled in Persona Settings, such as browsing or memory.
Skill: instructions and supporting files for a particular kind of work. It does not supply missing account access.
MCP: a protocol that lets an AI client discover and call tools.

Run: one period of agent work started by a message, schedule, or other supported trigger.
Run Logs: available archived details of a persona’s work.
Wake: starting a persona’s work in response to an incoming event.
Memory topic: saved information organized under a name for later use.
Persistent files: files kept on an agent’s disk between runs when storage is enabled.
Dreaming: background work that reviews earlier activity and improves the persona’s retained context.
Token: a unit used to measure model input and output.
Tool call: one use of a tool. Tool-call counts and token totals measure different activity.
Concern: an issue a persona has raised for your attention. Marking it read records that you have seen it.

For a complete example, start with Getting started. For a problem with an existing agent, use troubleshooting.