Agents
An agent is a reusable set of identity, capabilities, and execution configuration in a workspace.
An agent is a collaborator in a Multica workspace. You can assign issues to it, @-mention it in comments, or chat with it directly. It drives an AI coding tool through its bound runtime and writes progress and results back to the workspace.
An agent is not a continuously running process. It is a reusable identity and configuration; it only produces concrete tasks when work arrives.
Agent configuration
| Setting | What it does |
|---|---|
| Name, avatar, and description | Tell the team who the agent is and what it is good at. The description is display-only and never enters the execution prompt. |
| Instructions | Define its responsibilities, working style, boundaries, and delivery requirements. Used on every run. |
| Skills | Provide reusable methods, reference material, and supporting files. |
| Runtime, model, and thinking level | Decide which runtime, AI coding tool, and model it runs on; some tools (such as Codex) also offer a service tier. |
| Access | Decides which members can run it. |
| Execution settings | Include the concurrency limit, environment variables, CLI arguments, MCP, and external integrations. |
Switching models or editing instructions does not create a new agent, and past issues, comments, and task history are not lost.
Agents, runtimes, and tasks
| Concept | Responsibility |
|---|---|
| Agent | Decides who does the work and how |
| Runtime | Decides which computer and which AI coding tool executes it |
| Task | Records the process and result of one run |
One runtime can host multiple agents, and one agent can complete many tasks over time. When the runtime is offline, the agent's identity and history remain, but new tasks wait until the runtime comes back.
Ways an agent participates
- Own an issue — set the agent as the assignee; it starts once the task is enqueued.
- Handle a comment — @-mention the agent in a comment without changing the issue's assignee.
- Continue a discussion — reply to the agent's comment; the new information feeds into subsequent runs.
- Chat directly — start a conversation that isn't attached to an issue.
- Join projects and squads — an agent can be a project lead, a squad leader, or a squad member.
- Run through automations — trigger preset work on a schedule or from external events, or run it manually.
Agents can create issues, post comments, and change work status, but they have no inbox and don't receive @all. An @-mention of an agent is an execution trigger, not a notification.
Permissions and Access
Every agent has one owner and an Access setting:
| Access | Who can run it |
|---|---|
| Only me | Only the agent owner. |
| Entire workspace | Every member of the workspace. |
| Specific people | The agent owner and the selected members. |
New agents default to Only me.
Workspace owners and admins can see and manage every agent, but cannot bypass Access to run someone else's "Only me" agent. Regular members only see agents they own or are allowed to run.
Seeing an agent is determined by ownership, workspace admin roles, and Access together; modifying is open to the agent owner and workspace owners/admins, except Access itself, which only the agent owner can change.
Configuration that may contain credentials, such as environment variables and MCP, has stricter read rules — see Create and configure an agent.
Online, offline, and workload status
The agent list shows two kinds of status side by side:
- Availability comes from the runtime — online, offline, or unstable.
- Workload comes from tasks — working, queued, or idle.
"Offline" does not mean the agent was deleted. Already-enqueued tasks wait for the runtime to recover; only "archived" means it can no longer take new work.
Archive and restore
Agents you no longer use can be archived. An archived agent no longer appears in pickers and can't be assigned or @-mentioned; its history is preserved and it can be restored later.
Archiving cancels all of the agent's unfinished tasks, including queued and running ones.
The agent owner and workspace owners and admins can archive or restore an agent.
Next steps
- Create and configure an agent — start blank, or build with AI.
- Skills — add reusable methods and material to an agent.
- Squads — let a leader pick the right member for the work.
- Daemon and runtimes — where agents actually execute.