Privacy
This plugin sends user questions and (optionally) retrieved content to the configured LLM and embedding providers. It does not send any data to the plugin authors.
Operators are responsible for reviewing the privacy practices of their chosen providers.
What is sent, and to whom
- User questions → your configured LLM provider (for chat completion).
- Retrieved content (chunks from your knowledge base) → your configured LLM provider (as RAG context).
- Text to be embedded (queries and ingested documents) → your configured embedding provider.
All requests go directly from your WordPress server to the providers you configure. No telemetry, no call-home, no analytics.
What is stored
In your own WordPress database (prefixed {$wpdb->prefix}openrag_):
documents,chunks— your ingested knowledge base and its embeddings.chats,chat_sessions— conversation history and feedback. Each session has a random ownershipsecret; the secret is given only to the visitor who started that chat, and is required to view or modify that session’s history.mcp_servers— MCP server configuration and cached tool lists.
Provider API keys you enter in settings are also stored in your WordPress database (as options) and are sent only to the provider they belong to.
Removing data
Uninstalling is safe by default — your data is kept. Enable Advanced → Remove data on uninstall in settings to drop all tables and options when the plugin is deleted.
Provider terms and privacy policies
This plugin is self-hosted: requests go directly from your server to the provider you configure. Connecting to any provider is optional. Please review each provider’s terms and privacy policy:
- OpenAI (LLM/embeddings,
api.openai.com) — Terms of use · Privacy policy - Anthropic Claude (LLM,
api.anthropic.com) — Terms · Privacy - Groq (LLM,
api.groq.com) — Terms of use · Privacy policy - Cloudflare Workers AI (LLM/embeddings,
api.cloudflare.com) — Website terms of use · Privacy policy - Cloudflare Vectorize (vector storage,
api.cloudflare.com) — uses the same Cloudflare terms and privacy policy as above. - Ollama (local, runs on your own server) — Documentation. No data leaves your host.
- OpenAI-compatible / custom endpoints — if you point the plugin at a custom OpenAI-compatible endpoint, data is sent to the base URL you configure; consult that provider’s terms and privacy policy.
- MCP servers — if enabled, the plugin connects to the MCP server URL(s) you configure and may forward questions/context as tool-call arguments.