Let your agent wire up PicX
Paste one line into Claude Code, Cursor, Codex or any agent with shell access. It installs the SDK, wires your key, and proves the integration works with a real generation.
The line
What it does
- Reads the current API referenceFetches /llms.txt rather than relying on training data, so parameter names are the ones this API accepts today.
- Asks you for an API keyThe one step it cannot do alone — keys are created in a signed-in browser session. It writes the key to your env file and checks that file is gitignored.
- Installs the right SDKpicx-ai for JavaScript, picx-ai for Python, chosen from what your project already uses.
- Writes and runs a real generationNot a mock. It executes the call, and reports which step failed instead of claiming success if it does.
Before you run it
The prompt asks your agent to fetch and execute a remote file, which is worth reading first — it is plain text at /agent-setup/prompt.md. It installs one package, writes one env var, and makes one API call. It never asks for your key in chat, and it does not deploy anything.
You need a key first — create one on the API page. Prefer to do it by hand? The quickstart is the same steps, written for a human.