# Agent setup (https://upapi.io/docs/agent-setup)



Copy this sentence into any AI coding agent. It fetches https://upapi.io/docs/agent-setup/prompt.md and sets everything up itself:

```text
Fetch and execute the appropriate instructions to set me up for upAPI from https://upapi.io/docs/agent-setup/prompt.md
```

- [Claude Code](https://upapi.io/docs/agent-setup/claude-code)
- [Codex](https://upapi.io/docs/agent-setup/codex)
- [Cursor](https://upapi.io/docs/agent-setup/cursor)
- [OpenCode](https://upapi.io/docs/agent-setup/opencode)
- [VS Code](https://upapi.io/docs/agent-setup/vscode)

## What the agent does [#what-the-agent-does]

The sentence points the agent at
[prompt.md](https://upapi.io/docs/agent-setup/prompt.md), a plain Markdown file
written for agents. Following it, the agent:

1. finds your API key in the environment, or asks you for one once and stores it
   in a git-ignored `.env`;
2. connects the upAPI [MCP server](https://upapi.io/docs/mcp) to itself, preferring the hosted server,
   which signs in through your browser;
3. installs the [TypeScript SDK](https://upapi.io/docs/sdk) if the project has a `package.json`;
4. verifies the key and the MCP connection with two calls that cost zero units;
5. tells you what it did and what is left for you.

You are asked only for what an agent cannot do: a browser sign-in, or pasting a
key.
