Glossary
Words used across these docs, in plain terms.
| Term | Meaning |
|---|---|
| A2A | Agent-to-agent: one agent handing a task to another, on the same machine or another one. See A2A. |
| Activity | The recorded history of work and routing decisions. See Activity. |
| Agent | An AI worker with a workspace, instructions, and a model. |
| Channel | A connection that carries messages in from, and replies out to, a tool such as Telegram or GitLab. See Channels. |
| Cron | The timing format behind a schedule, such as 0 9 * * 1-5 for 9:00 on weekdays. In agentx.json, schedules live under crons. |
| Daemon | The AgentX background service that receives work and runs agents. |
| Dashboard | The separate browser server for setup and inspection. |
| Jev | An optional decision model that answers typed questions (yes/no, pick one of these) for AgentX. The agent's main model still does the open-ended work. See Jev and typed decisions. |
| MCP | Model Context Protocol, a standard way for AI tools to call outside tools. agentx serve --stdio offers AgentX to editors this way. |
| Mesh | Connected machines that can hand work to one another. |
| Node | A machine running AgentX. |
| Peer | Another node that this node is connected to in the mesh. |
| Provider | The company or tool that supplies the model, such as an API key account or a signed-in CLI. |
| Review | The check AgentX runs after a task finishes to write down what is left to do. Its results appear in Monitor. |
| Run | One execution of a task or workflow. |
| Schedule | A timed trigger for a task. |
| Seat | One kind of typed question AgentX can hand to a decision model, such as "which on-screen control matches this description?". |
| Tier | An agent's engine: claude-code, codex-cli, opencode, sdk (Anthropic API key) or orchestrator (any provider). The dashboard calls it AI engine. |
| Workflow | A saved automation with steps. See Workflow schema. |
For commands and accepted configuration, see the CLI and configuration references.
Check it worked
- Browser: open
/glossaryon your dashboard. It has its own in-app glossary for the words shown on screen. - Every term used on a docs page links here or is explained where it first appears.
If something is wrong
- A word on a page isn't explained here or on that page: open an issue naming the page and the word.
- A definition here disagrees with a page: the page about that feature is the one to trust; report the mismatch.