Skip to main content

LibreChat

Features​

  • Multi-Provider Support: OpenAI, Azure, Anthropic, Google, and custom endpoints
  • Plugin System: Extensible with custom plugins and integrations
  • Conversation Management: Organize chats with folders and search functionality
  • User Authentication: Secure multi-user environment with role-based access
  • Model Switching: Switch between different AI models within conversations
  • File Uploads: Support for document analysis and image processing
  • Custom Presets: Save conversation settings and system prompts
  • Message Export: Download conversations in various formats

Getting Started​

Generate password hash​

Replace SUPER_SECURE_PASSWORD with your password:

docker run --rm node:20-alpine sh -c 'mkdir /app && cd /app && npm init -y >/dev/null && npm install bcryptjs >/dev/null && node -e "console.log(require(\"bcryptjs\").hashSync(\"SUPER_SECURE_PASSWORD\", 10))"'

Add a user via MongoDB​

docker exec -it librechat-mongodb mongosh
use LibreChat
db.users.insertOne({
  email: "admin@example.com",
  password: "<PASTE_HASH>",
  role: "ADMIN",
  emailVerified: true,
  createdAt: new Date(),
  updatedAt: new Date()
})

Add Google Search tool​

Fill out these .env values following the LibreChat docs:

GOOGLE_SEARCH_API_KEY=
GOOGLE_CSE_ID=

Add agents​

Create agents (with "file search" capability) for use with Windmill scripts. After creating, the agent_id appears at the top of the side panel.

Debugging​

# View logs
docker compose exec librechat-api sh -lc 'cat /app/api/data/logs.json'

# View violations/bans
docker compose exec librechat-api sh -lc 'cat /app/api/data/violations.json'

# Find MongoDB user ID
docker exec -it librechat-mongodb mongosh
use LibreChat
db.user.find()