Setyenv · Docs
  • English
  • Spanish
  • Chinese
  • Japanese
  • Arabic
  • German
  • French
  • Hindi
  • Indonesian
  • Italian
  • Dutch
  • Portuguese
  • Russian
  • Turkish

Getting started

WP-PFAgent home

From zero to a first agent conversation in about 5 minutes.

Prerequisites

  • WordPress 6.5 or newer.
  • PHP 8.1 or newer.
  • An API key for at least one LLM provider (OpenAI, Anthropic, Gemini, etc.).
  • Optionally, the Setyenv suite (WP-PFManagement / WP-PFWorkflow) if you also want the data-modelling and workflow-automation surfaces. WP-PFAgent works on its own without them — it manages WordPress core and your popular plugins directly.

1. Install

WP-PFAgent is free. Install it like any WordPress plugin — from the WordPress.org plugin directory (Plugins → Add New, search for "WP-PFAgent"), or by uploading the plugin zip via Plugins → Add New → Upload Plugin → Install → Activate.

A new sub-menu appears under Setyenv → Agent. There is no licence key to enter.

2. Add an LLM provider

Setyenv → Agent → Providers → Add Provider:

  1. Pick a preset from the dropdown (e.g. OpenAI — GPT-4o).
  2. Paste your API key. The form may ask for extra fields depending on the provider (organisation id for OpenAI, deployment id for Azure, base URL for OpenAI-compatible).
  3. Click Test connection — sends a tiny health-check completion. Success means the key works and the agent can use this provider.
  4. Save.

You can add multiple providers — when starting a chat session you pick which one to use for that session.

3. Start a chat session

Setyenv → Agent → Chat:

  1. Pick a provider + model from the dropdown.
  2. Type your first message. Try something simple to confirm the pipe works: "List my workflows.".
  3. Press Enter.

The agent calls the workflow.list tool and shows you the list. That confirms: provider → agent → Workflow engine all wired.

4. Try a build instruction

Now ask something that BUILDS:

"Build a workflow that, when a WooCommerce order is over 200 EUR,
sends a personalised thank-you email using OpenAI."

The agent will:

  1. Draft the workflow as a JSON definition.
  2. Show you the draft inline (graph view + JSON view).
  3. Pause at the side-effect confirmation gate because creating a workflow is a write.
  4. Show two buttons: Approve and Reject (with notes).

Click Approve. The agent calls workflow.create with the drafted definition. A new workflow appears in Setyenv → Workflows, ready to test and activate as you would with any hand-built workflow.

What to do next

  • Read Side-effect confirmation gate — it is the single most important concept to understand before using the agent for real work.
  • Read Tools for the full list of what the agent can do.
  • Configure rate limits + budgets in Setyenv → Agent → Settings → Limits.