- Get started
- Best practices
- Capabilities
- Skills
- CLI
- Examples
- Examples overview
- Recipe: build an agent
- Recipe: automate with RPA
- Recipe: enter data into a web app
- Recipe: extract data from a desktop app
- Recipe: refactor and test a workflow
- Recipe: a queue-based process with REFramework
- Recipe: build a Maestro Flow
- Recipe: build a Maestro Case
- Recipe: coded app and API workflow
- Recipe: build a coded function
- Recipe: verify a release with Test Manager
- Recipe: extract data from documents with IXP
- Advanced
- Help
UiPath agents built with your coding agent, the skill that powers them, and where to go deeper.
UiPath agents are AI-driven automations that reason over a goal and take actions to reach it, rather than following a fixed sequence of steps. With a coding agent, you can describe the agent you want and have it scaffolded, wired up, and ready to run.
What you build
A typical agent handles work that benefits from judgment — triaging requests, drafting responses, or deciding which downstream automation to call. Your coding agent can build these in Python — with LangGraph, LlamaIndex, or OpenAI Agents — or in a low-code form, depending on what you ask for.
When an agent is one step in a larger process — with approvals, branching, and other systems — it can run as a node inside a Maestro Flow, which orchestrates it alongside people, RPA, and other automations.
The skill behind it
This work is powered by the uipath-agents skill, which your coding agent selects when you describe an agent. You can see it alongside the rest in the Skills catalog.
A typical starting prompt
A first request might look like this:
"Create a UiPath agent that reads a new support ticket, classifies its urgency, and drafts a suggested reply for a human to approve."
From there you refine the agent conversationally — adjusting its inputs, tools, and approval steps.
More prompts to try
- Build a UiPath agent that classifies incoming support tickets and posts the category to Slack.
- Build a Python agent that reads invoices from a queue, extracts fields, asks a human to confirm low-confidence ones, and writes the results back.
- Add a human-in-the-loop step that creates an Action Center task and resumes when it's completed.
- Set up an evaluation set for my agent with sample inputs and expected outputs.
Go deeper
For the full picture of UiPath agents — concepts, capabilities, and platform behavior — see the UiPath Agents documentation.