Glossary
Simple definitions for the terms you will see throughout AgentVibe. No jargon, no assumptions.
A
Agent Your AI assistant. It runs on your computer (or in the cloud), remembers your conversations, and can be reached through apps like Telegram, Discord, or the AgentVibe desktop window.
API Key A password-like string that lets your agent use an AI provider's models. For example, an Anthropic API key lets your agent use Claude. You get these from the provider's website.
Autopilot AgentVibe's health monitoring system. It periodically checks your agents, finds issues, and can automatically fix safe problems for you.
C
Channel A messaging platform your agent can communicate through. Telegram, Discord, WhatsApp, Slack, Email, and the AgentVibe desktop app are all channels. You can connect multiple channels to one agent.
CLI Command-Line Interface. A way to use AgentVibe from the terminal instead of the desktop app. Power users and automation scripts use this. Most people never need it.
Connector A small program installed on a remote server that lets AgentVibe deploy and manage agents there. Used when you deploy to your own VPS.
Credential Store Where AgentVibe keeps your API keys safe. On macOS it uses the Keychain, on Windows it uses the Credential Manager, and on Linux it uses an encrypted file.
D
Deploy Setting up and starting your agent. You pick a template, add your API key, and AgentVibe handles the rest -- creating the container, configuring the personality, and starting the agent.
Docker Software that runs applications in lightweight containers. AgentVibe uses Docker to run your agents in isolated environments. You install Docker once and AgentVibe handles everything else.
Dry Run A preview of what will happen without actually doing it. When you deploy, AgentVibe first shows you the plan and waits for confirmation before executing.
F
Framework The engine that powers your agent. Think of it like choosing between iPhone and Android -- both make calls, but they work differently. AgentVibe supports OpenClaw (most popular), Hermes (most channels), and any Docker-based agent.
Finding A security observation about your agent. Findings have severity levels (low, medium, high, critical) and some can be fixed automatically.
G
Gateway The network endpoint through which your agent receives messages. AgentVibe sets this up automatically during deployment.
H
Hardening Improving your agent's security. AgentVibe can scan for issues (like exposed ports or missing authentication) and fix safe ones automatically.
Host Where your agent runs. Options: your computer (free, default), a cloud provider (Fly.io, Hetzner, Railway), or your own server.
Hot-Reload When you edit your agent's personality file and the changes take effect immediately, without restarting anything.
I
Ingress A way to make a local agent reachable from the internet. Needed when Telegram or Discord webhooks need to reach your agent. AgentVibe supports Tailscale, ngrok, and Cloudflare tunnels.
M
Memory What your agent remembers between conversations. Stored in a file called MEMORY.md. Your agent updates it as it learns about your preferences and projects.
Model The AI brain powering your agent. Examples: Claude Sonnet, GPT-4o, Gemini. You can switch models anytime without redeploying.
P
Parrot AgentVibe's built-in AI guide that lives in the desktop app. Parrot helps you deploy agents, configure channels, switch models, and troubleshoot issues using natural language.
Persona Your agent's personality -- how it talks, what it knows, and what it does. Defined in a file called SOUL.md that you can edit in the app.
Provider A company that offers AI models. Anthropic (Claude), OpenAI (GPT), and Google (Gemini) are providers. You need an API key from at least one provider to use AgentVibe.
R
Registry AgentVibe's local database that tracks all your deployed agents. Stores their status, configuration, and history.
S
Skill An extra capability for your agent beyond basic conversation. Web search, file access, code execution, and memory management are examples of skills. You can add and remove them anytime.
SOUL.md The file that defines your agent's personality. Written in plain text. Edit it and your agent's behavior updates immediately.
T
Template A ready-to-use agent personality -- like an app in an app store. Templates come with a name, personality, model selection, channel recommendations, and scheduled tasks. Pick one and deploy.
W
Workspace The collection of text files that define your agent's personality, memory, and behavior. The most important file is SOUL.md (personality), but there are also files for identity, memory, startup instructions, and more.