Documentation / Settings

Settings

Current settings layout (Interface, Agents, Providers, Users, Security, Advanced)

Interface

Open Agent Builder → Settings → Interface. This tab controls how chat looks and how assistants address people. Site knowledge and standing guidance live under Knowledge; per-assistant personality is under Knowledge → Instructions.

Settings Interface tab with Basic/Advanced mode, addressing fields, and chat theme previews
Interface — mode, addressing, theme previews, font, and Getting Started toggle

Interface mode

  • Basic — everyday settings only (Interface, Agents, Providers, Users, Security).
  • Advanced — also shows developer tools: Skills, APIs, and Endpoints.

How assistants address people

Optional labels for administrators and frontend visitors. Leave blank to use each person’s WordPress display name.

Chat theme

Pick a visual theme with live preview cards. Applies to admin chat and frontend shortcodes.

  • Light (default on new installs) — clean white with WordPress blue accents.
  • Dark — deep purple dark theme.
  • Midnight — pure dark with emerald accents.
  • Ocean — deep blue with teal highlights.

Also set Chat font, whether to use the theme accent colour, and whether the dashboard shows the Getting Started checklist (toggle at the bottom of this tab).

Chat capabilities (audio, vision, white-label) are on the Agents tab — not Interface.

More detail: Chat styles and themes.

Agents

Site-wide chat features, compact provider/model overrides for active assistants, and the emergency stop. Set the site default provider on the Providers tab; theme is under Interface.

Settings Agents tab with chat features toggles and active agent provider overrides
Agents — chat features, per-assistant overrides, Disable All Agents

Chat features

  • Audio input — microphone / browser speech recognition.
  • Text-to-speech — read responses aloud (may use prepaid credits).
  • Vision / image input — attach images when the model supports vision.
  • White-label branding — hide “Powered by Agent Builder” (Pro).

Active agent overrides

Each row starts on the site default provider and model. Change only assistants that need a different setup. Dropdowns list connected providers only.

Disable All Agents

Emergency stop: deactivates agents, cancels jobs, and disconnects providers.

Providers

Connect AI providers and choose which connected provider is the site default. Only providers with a green status (connected) can be set as default.

Settings Providers list with status, models, Edit and Set Default actions
Providers — status, models, Edit, and Set Default for connected providers
  • Use Edit to add or update API keys and models.
  • Click Set Default on a connected row (green badge shows the current default).
  • Built-in options include Agentic AI, OpenAI, Anthropic, Google Gemini, xAI, Mistral, Meta Llama, Cohere, DeepSeek, Kimi, Ollama, and custom OpenAI-compatible endpoints.

Full guide: Manage LLM providers.

Users

Role matrices for plugin administration and AI agent access, plus daily query/token quotas. Hover the ? next to each permission for a short explanation. Administrators always retain full access and unlimited usage.

Settings Users tab with plugin administration and AI agent privilege matrices
Users — privileges, tooltips, and daily limits (rate limits are under Security)
  • Plugin Administration — who can open dashboard, agents, tools, audit log, settings, and run tasks.
  • AI Agents — frontend chat, admin bar chat, view installed agents, premium agents; daily queries/tokens per role and for visitors.

Request rate limits (per-minute throttle) live on Settings → Security, not Users.

Full guide: User roles and privileges.

Security

Default agent mode, message scanning, chat consent, retention, and request rate limits.

Settings Security tab with agent mode, scanning, retention, and request rate limits
Security — mode, scanning, consent, retention, and per-minute rate limits
  • Default agent mode — Supervised (approvals for write actions), Autonomous, or Disabled.
  • Message scanning — filter prompt injection and known attack patterns.
  • Require chat consent — optional notice before chat starts.
  • Conversation / audit log retention — days to keep local history (default 30).
  • Request rate limits — authenticated and anonymous requests per minute (moved here from Users).

High-risk tool calls still follow Approvals. Deep guide: Important security settings.

APIs

Advanced only (Interface mode = Advanced). Third-party keys used by specific tools — for example Google PageSpeed Insights for SEO tooling. Everyday LLM keys belong on the Providers tab.

Settings APIs tab for third-party service keys
APIs — optional third-party keys for tools

See also Connecting an AI provider.

Endpoints

Advanced only. REST and integration endpoints for custom clients and automations. Prefer connected providers on the Providers tab for day-to-day chat.

Settings Endpoints tab for advanced API endpoints
Endpoints — advanced integration surfaces

License

When Pro is installed, enter your license key (format AGNT-XXXX-XXXX-XXXX-XXXX) to unlock marketplace, Vector Store, white-label, channels, and priority support. Free Agent Builder does not require a key.

Details: Licensing and pricing.

Instructions and Memory are under Knowledge (not Settings). MCP lives under Connectors when Pro is available.

Get Support →