Using Skills: Web Search, Image Gen & Deep Research

Skills

Skills extend what your agents can do beyond reading and writing code. With skills enabled, your agents can search the web, generate images, conduct multi-step research, and connect to external services.

---

Enabling Skills

Skills are enabled per-agent. When you create an agent, specify which skills it should have access to:

TypeScript

Python

Or toggle them in the UI when creating or editing an agent.

---

Web Search

With web search enabled, your agent can look up real-time information:

The agent will search the web, extract relevant information, and synthesize it into a response — with citations.

---

Image Generation

Your agent can create images on demand:

Generated images are saved to the agent's workspace and can be downloaded or used in follow-up tasks.

---

Deep Research

Deep research is a multi-step skill. The agent plans a research strategy, executes multiple web searches, cross-references sources, and produces a comprehensive report:

This produces a structured report with citations, often running 2,000+ words.

---

Custom MCP Servers

For integrations beyond built-in skills, connect Model Context Protocol servers:

TypeScript

Python

With MCP servers, your agent can interact with GitHub, Notion, Google Drive, Slack, databases, and any service that exposes an MCP endpoint.

---

Combining Skills

The real power comes from combining skills. An agent with web search, deep research, and image generation can:

1. Research a topic in depth
2. Generate supporting visuals
3. Write a complete blog post with images
4. Save everything to the workspace

All from a single prompt.