Skip to content

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:

SettingDescription
Share ButtonShare entire conversations
Message ShareShare individual messages
Message FeedbackThumbs up/down on responses
Message StatsToken usage and timing info
Pin ButtonPin conversations to top
Create ButtonNew conversation button
Assistant DropdownSwitch assistants mid-chat
Agent ThinkingShow reasoning traces
Chat SuggestionsStarter questions on new chat

Follow-up questions

SettingDescription
Generate Follow-upsAuto-suggest next questions after responses
Question CountHow many to generate (1-5)

INFO

Follow-up generation uses an additional LLM call per response. Consider the cost impact.

SettingDescription
User RegistrationEnable/disable new signups
Terms & Privacy LinksShow legal documents on signup form

WARNING

Only enable terms and privacy links if you have proper legal documents set up.

Conversation titles

SettingDescription
Auto-generateUses first message as title
AI ModelSelect 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

SettingDescription
TemperatureCreativity/randomness (0-2)
Max tokensResponse length limit
Top PNucleus sampling parameter
Other LiteLLM paramsModel-specific settings

RAG settings

SettingDescription
Retrieval countNumber of chunks to retrieve
Similarity thresholdMinimum score for inclusion
Search typeSimilarity or MMR
Multi-queryEnable query expansion
RerankingRe-score top candidates
Hybrid searchBM25 + 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:

FieldDescription
NameAPI model identifier
Display nameHuman-readable name
ProviderWhich provider hosts this model
CapabilitiesChat, embedding, vision, etc.
Token limitsContext and output limits
StatusActive, inactive, or deprecated

Knowledge bases (/admin/kbs)

ActionDescription
CreateNew KB with embedding model and chunk settings
UploadAdd documents via drag-and-drop
ConfigureAdjust chunking, embedding model
ReindexRegenerate all embeddings
DeleteRemove KB and all documents

User management (/admin/users)

ActionDescription
View usersList all users with activity info
Activate/deactivateControl user access
Grant superuserPromote to admin
Revoke superuserRemove admin privileges

TIP

The first registered user automatically becomes a superuser.

OAuth providers (/admin/oauth-providers)

Configure social login:

  1. Google — Client ID, secret, scopes
  2. GitHub — Client ID, secret
  3. Microsoft — Client ID, secret
  4. 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

MIT License