Agents Settings
Agents Settings is the largest section of App Settings — but it's app-wide toggles and plan-gated features, not where you configure an individual agent. For a specific agent's persona, autonomy mode, provider, or capabilities, see the Agents guide instead; this page is the five subsections that sit above every agent: 3D Room, Session Replay, mesh and provider syncing, orchestration, and a terminal visual effect.
3D Room
Controls for NEXUS's 3D renderer of the Nexus Room — the shared space where agents with a live session appear together. 3D Room requires a paid plan. Without one, this subsection shows a "Not entitled" status and an upgrade link instead of the controls below.
| Label | What it does | Default |
|---|---|---|
| Status | Whether 3D Room is active, and whether your plan includes it. | Inactive |
| Activate 3D Room / Deactivate 3D Room | Turns the 3D renderer on or off. | Off |
| Download Asset Bundle / Update | Downloads or updates the asset bundle 3D Room needs before it can render. | Not installed |
| Renderer | 2D or 3D, once 3D Room is active. | 2D |
| Movement speed | Camera movement speed in 3D. | 5.0 m/s |
| Mouse sensitivity | Camera turn sensitivity in 3D. | 0.3°/px |
| Field of view | Camera field of view in 3D. | 75° |
| Invert Y axis | Inverts vertical look in 3D. | Off |
| Start in third-person mode | Starting camera mode in 3D. | Off |
| Head-bob | Camera head-bob amount in 3D. | 0.04 m |
2D is the default renderer and available to everyone regardless of plan; the camera controls above only apply once 3D Room is both entitled and activated.
Session Replay
| Label | What it does | Default |
|---|---|---|
| Enable Session Replay | Turns on an experimental view that replays which files a past session touched, as a static map of the repository it ran in. | Off |
Session Replay is experimental and off by default. It's local-only, has nothing to do with a session's live status, and is a separate, later thing you turn on once you have sessions worth reviewing — see Running & monitoring for what live status means while a session is actually running.
Mesh, Providers & Team Expansion
| Label | What it does | Default |
|---|---|---|
| Mesh Messaging | Turns on agent-to-agent mesh messaging. | On |
| Sync Agents to All CLI Providers | Keeps your agent roster in sync across every configured CLI provider. | Off |
| Agents Can Expand Teams | Lets an agent request inviting or creating another agent. | Off |
Any team-expansion request an agent makes always arrives as a message you see and choose to act on — it never happens silently, regardless of this toggle's state.
Orchestration
| Label | What it does | Default |
|---|---|---|
| Cross-Agent Orchestration | Lets agents compose — one agent triggering another, which can itself trigger, message, or write further — instead of only ever acting alone. | Off |
| Max chain depth | Caps how many hops a chain can run. | 2 (range 1–10) |
| Max agents per chain | Caps how many distinct agents one chain can involve. | 3 (range 1–10) |
| Require approval for each hop | Pauses for your approval before every hop, not only the first. | On |
| Allowed agent slugs (comma-separated) | The allow-list. Only agents named here can start a governed chain. | Empty |
| Stop All Orchestration | Immediately halts every agent conversation and orchestration chain, app-wide. | — |
Cross-Agent Orchestration is off by default, gated to plans that include it, and inert until you fill in the allow-list. Turning the master toggle on by itself unlocks nothing: an empty allow-list means no agent can start a chain no matter what else is configured here. Depth, per-chain agent count, and per-hop approval only start to matter once at least one agent is actually named in that list.
Visual Effects
| Label | What it does | Default |
|---|---|---|
| Terminal activity effect | A visual effect reflecting terminal output volume. | Off |
This effect is tied only to how much output a terminal is producing — it is never a measure of an agent's context usage or anything about what it's actually doing.
Where to go next
- Settings reference — the control-by-control contract for every subsection on this page.
- Agents guide — configuring an individual agent's persona, autonomy, provider, and capabilities.
- Capabilities & skills — the per-agent mesh capabilities that Orchestration's allow-list sits above.