System Configuration
Access the admin panel at /admin/system-config to control system-wide settings. These are stored in the database and take effect immediately.
Chat UI controls
Toggle visibility of chat interface elements:
| Setting | Description |
|---|---|
| Share Button | Share entire conversations |
| Message Share | Share individual messages |
| Message Feedback | Thumbs up/down on responses |
| Message Stats | Token usage and timing info |
| Pin Button | Pin conversations to top |
| Create Button | New conversation button |
| Assistant Dropdown | Switch assistants mid-chat |
| Agent Thinking | Show reasoning traces |
| Chat Suggestions | Starter questions on new chat |
Follow-up questions
| Setting | Description |
|---|---|
| Generate Follow-ups | Auto-suggest next questions after responses |
| Question Count | How many to generate (1-5) |
INFO
Follow-up generation uses an additional LLM call per response. Consider the cost impact.
Authentication & legal
| Setting | Description |
|---|---|
| User Registration | Enable/disable new signups |
| Terms & Privacy Links | Show legal documents on signup form |
WARNING
Only enable terms and privacy links if you have proper legal documents set up.
Conversation titles
| Setting | Description |
|---|---|
| Auto-generate | Uses first message as title |
| AI Model | Select a model to generate smart titles |
When an AI model is selected, titles are generated from the first message using that model.
Assistant defaults
Set platform-wide defaults at /admin/assistant-defaults:
Generation settings
| Setting | Description |
|---|---|
| Temperature | Creativity/randomness (0-2) |
| Max tokens | Response length limit |
| Top P | Nucleus sampling parameter |
| Other LiteLLM params | Model-specific settings |
RAG settings
| Setting | Description |
|---|---|
| Retrieval count | Number of chunks to retrieve |
| Similarity threshold | Minimum score for inclusion |
| Search type | Similarity or MMR |
| Multi-query | Enable query expansion |
| Reranking | Re-score top candidates |
| Hybrid search | BM25 + dense fusion |
Individual assistants can override these defaults.
Models & providers
Managing providers (/admin/providers)
Add LLM providers with their API configuration:
- OpenAI — API key and base URL
- Anthropic — API key
- Google Gemini — API key
- Custom — Any OpenAI-compatible API
Managing models (/admin/models)
Configure available models:
| Field | Description |
|---|---|
| Name | API model identifier |
| Display name | Human-readable name |
| Provider | Which provider hosts this model |
| Capabilities | Chat, embedding, vision, etc. |
| Token limits | Context and output limits |
| Status | Active, inactive, or deprecated |
Knowledge bases (/admin/kbs)
| Action | Description |
|---|---|
| Create | New KB with embedding model and chunk settings |
| Upload | Add documents via drag-and-drop |
| Configure | Adjust chunking, embedding model |
| Reindex | Regenerate all embeddings |
| Delete | Remove KB and all documents |
User management (/admin/users)
| Action | Description |
|---|---|
| View users | List all users with activity info |
| Activate/deactivate | Control user access |
| Grant superuser | Promote to admin |
| Revoke superuser | Remove admin privileges |
TIP
The first registered user automatically becomes a superuser.
OAuth providers (/admin/oauth-providers)
Configure social login:
- Google — Client ID, secret, scopes
- GitHub — Client ID, secret
- Microsoft — Client ID, secret
- Custom — Any OAuth 2.0 provider
Set user mapping rules (email, name, avatar).
Analytics (/admin/analytics)
View usage statistics:
- Token usage trends
- Request counts
- Response latency
- Daily/weekly breakdowns
Tips
- Refresh frontend after changing public configs to see changes
- Test in incognito to verify behavior for new users
- Monitor analytics to optimize model selection and costs
- Keep terms hidden unless you have proper legal documents