Working with Agents
Build custom AI agents in Nebula for code, web, and app tasks. Choose from built-in agents or create your own to automate complex workflows.
Agents are AI assistants that handle different tasks. When you ask Nebula to do something, it acts as a super-agent — an intelligent coordinator that figures out which specialized agents are needed and delegates work to them. Nebula comes with built-in agents, public agents you can clone from Explore, and automatically created agents for connected apps.
Not sure how to configure your agent? Ask Nebula to do it for you — it has tools to create and modify agents directly.
Built-in agents
Writes and runs Python, Bash, or TypeScript on your Device. Common data science libraries like pandas, numpy, and matplotlib are pre-installed.
Searches the web, reads pages, extracts structured data, and gathers information from any public website.
Generates images from text descriptions, transcribes audio recordings, converts text to speech, and processes media files.
Sends and reads emails from your Nebula inbox at username@nebula.me.
Explore agents
The agent registry has public agents you can clone into your workspace and run instantly. Browse featured agents, filter by app, or start from popular workflows already published by the community.
Cloned agents become your own copy. You can edit them, connect your accounts, add triggers, or use them as-is.
See Agent Registry for cloning and publishing.
Auto-created agents
When you connect an app like Slack or GitHub, Nebula automatically creates a dedicated agent for it. Your Slack Agent can read channels, post messages, and manage your workspace. Your GitHub Agent can triage issues, manage pull requests, and more.
You don't need to configure these agents; they're ready to use as soon as you connect the app.
How agents get work done
For longer tasks, you can watch an agent's progress in real time in the task timeline view on the web app.
Agent details panel
Click any agent in the sidebar to open a DM, then click the agent name in the conversation header to open the details panel. Everything about the agent is in one scrollable view — name, model, description, goals, system prompt, triggers, tools, accounts, and variables. All editable inline.
See Configure & Manage Agents for the full guide on every setting.