Configure & Manage Agents
Configure your Nebula agent's model, system prompt, goals, tools, accounts, variables, and triggers — all from the details panel.
Every agent has a details panel where you can view and edit its configuration. Open it by clicking the agent name in a conversation header.
Opening the details panel
Everything below is shown in a single scrollable view.
Name
Click the agent's name in the hero area to rename it. Press Enter to save or Escape to cancel. The name helps Nebula decide when to delegate tasks to this agent, so keep it descriptive.
Model
Click the model badge next to the name (e.g., "Claude Sonnet 4.6") to open the model selector. Browse models by provider, capability, and cost — then click one to apply it immediately. The agent keeps all its other settings.
To let Nebula choose the best model automatically, click the reset button in the model selector. See Choosing a Model for guidance on when to pin a specific model.
Switch my research agent to GPT-4.1.
Done! Your research agent now uses GPT-4.1. All existing conversations and instructions are unchanged.
Description
Click the edit icon next to Description to update what the agent does. This is a short summary that helps Nebula understand the agent's purpose when deciding whether to delegate a task to it.
Goals
Goals are objectives you set for the agent — useful for tracking what the agent should focus on.
Click the edit icon next to Goals, then click Add goal. Enter a description and set the priority and status. Click Save when done.
Each goal has two properties:
| Property | Options |
|---|---|
| Priority | High, Medium, Low |
| Status | Active, Achieved, Paused |
Update these as the agent progresses toward its objectives.
System prompt
The system prompt defines the agent's role, behavior, and boundaries. It's sent alongside every message the agent receives.
Click the edit icon next to System Prompt. Modify the text directly in the editor — it supports full markdown. Click Save when done.
For long prompts, the panel shows a collapsed preview with a Show more button.
Ask Nebula to update it for you:
Update my Support Triage agent to always classify issues as P1-P4 and tag the relevant team.
Done — updated the system prompt to include priority classification (P1–P4) and team tagging for all incoming issues.
You don't need to describe the agent's tools in the system prompt — Nebula handles capability awareness automatically. Focus on role, tone, and domain knowledge. See System Prompts for best practices.
Triggers
Any automations linked to this agent appear here. Each trigger shows its name, schedule or event type, and whether it's active or paused.
Tools
Tools are the APIs and services the agent can call. They're organized by toolkit — expand any toolkit to see its capabilities and individual endpoints.
Toolkits group related capabilities under a single service (e.g., "Web Search", "Code Execution", "Slack"). Expand a toolkit to see what it can do.
To remove a toolkit from the agent, hover over it and click the ×.
Custom API tools show their endpoint details when expanded:
Accounts
Accounts are the third-party services connected to the agent (e.g., Slack, GitHub, Gmail). Each account shows its connection status.
A green checkmark means the account is active and working. No action needed.
When you have more than one account for the same service (e.g., two Slack workspaces), a dropdown appears. Select which account the agent should use.
Click Connect to authorize the service via OAuth. You'll be redirected to the app's authorization page.
To remove an account from the agent, hover over it and click the ×.
Variables
Some tools require credentials or configuration values — like an API token for Sentry or a key for a custom service. Variables show their status:
A green Set badge means the variable has a value and the agent can use it. Click the edit icon to update it (e.g., after rotating a key).
An orange Missing badge means the variable needs a value before the agent can work. Click Configure to set it. If the agent has missing variables, a configuration alert appears at the top of the panel.
Never paste secrets directly into a chat message. Always use the variable configuration form in the details panel.
Knowledge base
If the agent has a knowledge base, it appears as a collapsible markdown section near the bottom of the panel. This is reference material the agent can draw from when responding.
Deleting an agent
Scroll to the bottom of the details panel and click Delete Agent. You'll be asked to confirm — deletion is permanent. The agent's configuration is removed, but any conversations and work it completed are preserved in your history.
System agents (marked with a Platform badge) and the main Nebula agent cannot be deleted.
What's next?
Create Custom Agents
Build custom AI agents in Nebula to automate specific tasks, workflows, or app integrations. Create them from the sidebar or in chat.
Agent Tools
Explore the tools available to Nebula agents — web search, code execution, browser automation, app integrations, media processing, and more.