The memory runs in every organization. Everything else is a module you turn on, in the app under All modules or in your daemon’s daemon.toml.
| Module |
What it does |
| Memory |
Records, the graph, sessions, conversations, cards and documents; always on |
| Module |
What it does |
| Board |
Boards with tasks, their discussion and their decisions |
| Team |
Agents with roles that work on the board, and the trust you give them |
| Chief |
The organization’s main agent, over every module |
| Agent |
The harness’s own assistant: turns with a model and tools |
| Skills |
A library of instructions, each also an MCP prompt |
| Models |
The model providers your agents use: OpenRouter, Anthropic, Z.AI and OpenAI, with your own keys |
| Module |
What it does |
| Workflow |
Graphs of steps over every module’s tools |
| Scheduler |
Jobs at cron times or intervals |
| Webhooks |
Signed events from outside become tasks or workflow runs |
| Custom modules |
Your own modules in Python or TypeScript, run once a person approved their code |
| Module |
What it does |
| HTTP |
Outside APIs, with your own credentials kept as connections |
| Mail |
Mail with your own transport, from templates |
| Telegram |
Messages, approvals with buttons, and the chief in a chat |
| Module |
What it does |
| Observe |
Reads metrics, logs, alerts and a cluster’s state |
| Ops |
Changes a running system only through typed, approved actions |
A daemon also has the built-in tool modules: tools-fs (files), tools-shell (a shell and tmux) and tools-git, each off until you name it with what it may reach (an organization in the cloud gets them only from a daemon it connects as its runner). See Files, shell and git.
Your own module: Writing a module.
