Documentation / Settings

Settings

Updated February 28, 2026

Agent Builder → Settings contains eight tabs that control how the plugin behaves. Changes to each tab are saved independently — clicking Save on the General tab will not overwrite your License or Security settings.

General

This is the most important tab to complete before using the plugin.

  • AI Provider — choose from OpenAI, Anthropic, xAI (Grok), Google (Gemini), Mistral, or Ollama (local, free). The provider you select here becomes your default provider — all assistants without a per-assistant override will use it. You can also switch the default quickly from the Dashboard Status card. See Connecting an AI Provider for setup instructions for each.
  • API Key — paste your provider’s secret key. The key is stored encrypted in the database and never exposed in the admin UI after saving. Not required for Ollama.
  • Model — select a specific model for this provider (for example gpt-4o, claude-3-5-sonnet-20241022, or grok-3). Leaving this blank uses the provider’s recommended default.
  • Agent Mode — controls how assistants handle file-system changes:
    • Supervised (default) — code changes are queued for your review before they execute. Documentation updates happen automatically.
    • Autonomous — all actions execute immediately without an approval step. Use with care: in this mode an assistant can modify files, update settings, or change database records on your site without any further confirmation from you.
    • Disabled — assistants cannot make any file changes at all.

Agent Configuration

Below the global settings, the Agent Configuration table lets you override the AI provider, model, and mode for each active assistant individually. This gives you fine-grained control — for example, you can run your Content Writer on a fast, cheap model while keeping your Security Monitor on a more capable one, or disable a specific assistant without touching the global settings.

Only providers you have already configured (i.e. have a saved API key for) appear in each assistant’s provider dropdown. The Model dropdown is populated automatically when you choose a provider.

  • Provider — override the AI provider for this assistant. Leave blank to use the global provider.
  • Model — override the model for this assistant. Leave blank to use the global model.
  • Mode — override the agent mode for this assistant:
    • Global default — inherits the Agent Mode set above.
    • Supervised — changes for this assistant are always queued for your review, even if the global mode is Autonomous.
    • Autonomous — this assistant acts immediately without confirmation, even if the global mode is Supervised.
    • Disabled — this assistant will not respond to chat at all.
  • Click Reset on any row to clear all overrides for that assistant and return it to global defaults.

License

Enter your license key here (format: AGNT-XXXX-XXXX-XXXX-XXXX). After saving, the plugin contacts the licensing server and shows:

  • Status — Active (green), Expiring (yellow), Expired/Revoked (red), or Free
  • Type — Personal or Agency
  • Expiry date and Activations used / limit

Use the Deactivate button to free up an activation slot before moving the plugin to a different site. Purchase or renew licenses at agentic-plugin.com/pricing.

Cache

Response caching stores LLM responses so that identical prompts are not sent to the AI provider again — useful if multiple users ask the same question. Options:

  • Enable response caching — toggle the cache on or off.
  • Cache TTL — how long (in seconds) cached responses are kept. Default is 3600 (one hour).
  • Clear cache — immediately purge all cached responses.

Security

  • Enable security features — when on, all chat requests are validated and rate-limited.
  • Rate limit (authenticated users) — maximum requests per hour for logged-in users. Default: 30.
  • Rate limit (anonymous users) — maximum requests per hour for visitors. Default: 10.
  • Allow anonymous chat — when enabled, visitors who are not logged in can use the chat interface.

Permissions

The Permissions tab controls which actions assistants are allowed to perform. All scopes are disabled by default — you must explicitly enable each type of action you are comfortable with.

Available permission scopes include reading and writing files, updating WordPress options, managing transients, and modifying post meta. The Confirmation Mode setting determines whether proposals await your approval or execute immediately. See Permissions and Safety for a full breakdown.

Users

The Users tab lets you control which WordPress roles are allowed to interact with AI assistants. You can grant or restrict chat access on a per-role basis.

Styles

The Styles tab controls the appearance and feature toggles for the chat interface. You can select a Chat Theme (Dark, Light, Midnight, or Ocean), toggle speech input, image recognition, cost tracking display, and white-label branding on or off. See Chat Styles & Themes for full documentation.

Providers

The Providers tab opens the Manage LLM Providers interface where you can view, add, edit, and delete AI providers. Agent Builder ships with nine built-in providers, and you can register any OpenAI-compatible endpoint as a custom provider. See Manage LLM Providers for full documentation.

Browse all documentation →