ChatLima

Feature-rich MCP-powered AI chatbot with multi-model support and advanced tools.

Features • MCP Configuration • License

# ChatLima Feature-rich MCP-powered AI chatbot with multi-model support and advanced tools. This project is a fork of [scira-mcp-chat](https://github.com/zaidmukaddam/scira-mcp-chat) by Zaid Mukaddam, with extensive additional features and modifications by Garth Scaysbrook. ## Features ### 🤖 AI Model Support - **Dynamic Model Loading**: Real-time fetching of AI models from OpenRouter and Requesty APIs, eliminating manual model updates - **300+ AI Models**: Access to 300+ language models through dynamic provider integration including: - OpenAI GPT-4.1 series (Full, Mini, Nano) and GPT-4O series (via OpenRouter and Requesty) - Anthropic Claude 4 series (Sonnet, Opus), Claude 3.5, 3.7 Sonnet with thinking variants (via Anthropic and Requesty) - Google Gemini 2.5 Pro and Flash variants with built-in thinking capabilities (via OpenRouter and Requesty) - DeepSeek R1 models including DeepSeek R1 0528, V3, Chat, and Qwen3 8B variants - Meta Llama 4 Maverick and 3.1/3.3 70B series (via OpenRouter and Requesty) - Mistral Magistral Small and Medium 2506 models with thinking variants - X AI Grok 3 and 4 models with advanced reasoning and tool calling support - MiniMax M1 models with 456B parameters and extended context - Specialized models: TheDrummer Anubis 70B, Inception Mercury, Baidu ERNIE-4.5-300B, Kimi K2 - **Automatic Model Discovery**: Dynamic model discovery with smart caching and automatic updates - **Enhanced Model Metadata**: Rich model information including capabilities, pricing, and performance characteristics - **Intelligent Model Filtering**: Automatic filtering of blocked, deprecated, and unavailable models - **Dynamic API Key Management**: Use your own API keys for any provider with runtime overrides - **Smart Credit Validation**: Bypasses credit checks when using personal API keys - **Credit-Tier Model Pricing**: Per-message costs (1–30 credits) shown in the model picker with real-time balance checks - **Model Picker**: Real-time model availability and access validation with comprehensive descriptions and search functionality - **Provider Health Monitoring**: Comprehensive provider status tracking and health checks ### 🖼️ Image Input & Multimodal Support - **Complete Image Upload System**: Drag-and-drop interface with support for JPEG, PNG, and WebP formats - **Advanced Image Processing**: Intelligent validation, compression, and format optimization with 20MB file limit - **Multiple Image Support**: Upload up to 5 images per message for complex visual conversations - **Image Preview & Management**: Real-time previews, full-screen modal viewing, and metadata display - **Vision Model Integration**: Works seamlessly with all vision-capable AI models for image analysis, OCR, and visual question answering - **Detail Control**: Configurable image detail levels (low, high, auto) for cost and quality optimization - **Download Support**: Direct image download functionality from the modal viewer - **Smart UI Integration**: Seamless integration with existing chat interface and message flow ### 🎨 Comprehensive Presets System - **Custom AI Configurations**: Save, manage, and apply personalized AI model configurations - **Built-in Templates**: Ready-to-use presets for coding, writing, research, analysis, and specialized tasks - **Template Categories**: - **Coding**: Advanced Code Architect, DeepSeek V3 Code Expert, Kimi K2 Agentic Coder, GPT-4.1 Mini Rapid Coder - **Analysis**: DeepSeek R1 Deep Reasoner, Grok 4 Research Analyst, Gemini Pro 2.5 Data Scientist - **Writing**: Claude Sonnet 4 Technical Writer, Gemini 2.5 Flash Content Creator - **General**: Executive Assistant, Personal Assistant, Strategic Problem Solver, Learning Tutor - **Visual Management**: Intuitive preset selection with indicators, tooltips, and responsive design - **Default Preset Support**: Smart default handling with proper constraint management - **Template-Based Creation**: Quick preset creation using predefined templates - **Contextual Hints**: Template name hints during preset creation for better UX ### 🔐 Authentication & User Management - **Google OAuth Integration**: Seamless sign-in with Google accounts - **Anonymous Users**: Support for anonymous usage with unique tracking (10 messages/day) - **Flexible Message Limits**: - Anonymous users: 10 messages/day - Free Google users: 20 messages/day - Monthly subscribers: ~1,000 credits/month; each message costs 1–30 credits depending on model - Yearly subscribers: ~12,000 credits/year; each message costs 1–30 credits depending on model - **Enhanced Daily Message Tracking**: Tamper-proof daily message usage tracking that prevents limit bypass vulnerabilities - **Automated User Cleanup**: Intelligent system to remove inactive anonymous users for optimal database performance - **Admin Cleanup Dashboard**: Comprehensive interface with preview, configuration, and monitoring capabilities - **Credit System**: Integrated billing and credit management with Polar (for monthly plan users) - **Better Auth Integration**: Modern authentication with session management - **Robust Credit Management**: Enhanced validation to prevent negative credit balance issues - **Subscription Type Detection**: Automatic detection of monthly vs yearly subscriptions for proper access control ### 💳 Billing & Payment System - **Flexible Subscription Plans**: Choose the plan that fits your usage pattern - **Monthly Plan ($9/month)**: ~1,000 credits per month with access to the full model catalog - **Yearly Plan ($90/year)**: ~12,000 credits per year with the full model catalog — save ~17% vs monthly - **Tiered Credit System**: Variable credit costs based on model pricing (shown as 1c–30c in the model picker) - **Economy (1 credit)**: Standard and OpenRouter `:free` models - **Standard (2 credits)**: Mid-tier frontier models - **Pro (5 credits)**: High-capability models - **Frontier (15 credits)**: Very high-cost models - **Ultra (30 credits)**: Top-tier reasoning models (e.g., o1-pro, o3-pro) - **Transparent Pricing**: Credit costs displayed in model picker for informed decision-making - **Smart Credit Validation**: System automatically checks if you have sufficient credits before allowing expensive model usage - **Polar Integration**: Complete integration with Polar billing platform for customer management - **Production Checkout System**: User-friendly purchase flow with smart user handling and seamless sign-in - **Upgrade Page**: Beautiful side-by-side plan comparison with clear benefits and pricing - **Customer Portal Access**: Direct access to Polar customer portal for subscription management - **Credit Purchase Workflow**: Streamlined process for purchasing AI usage credits (for monthly plan users) - **Subscription-Based Access Control**: Intelligent model filtering based on subscription type - **Access gating**: When `BILLING_ENFORCED=true`, chat requires subscription or BYOK (see `lib/config/access-policy.ts`) - **Paid Web Search**: Premium web search feature with credit-based billing and usage tracking (5 credits per search) - **Comprehensive Error Handling**: Dedicated error pages for failed, canceled, and problematic transactions - **Environment-Based Configuration**: Secure Polar server environment selection ### 🛠️ Model Context Protocol (MCP) - **Latest MCP 1.13.0 Support**: Full compatibility with the latest Model Context Protocol specification - **Enhanced Protocol Headers**: Proper MCP-Protocol-Version headers for HTTP transport - **Multiple Transport Types**: Support for SSE, stdio, and HTTP Streamable connections - **OAuth 2.1 Authorization**: Built-in OAuth 2.1 support for remote MCP servers requiring user authentication - **Secure Token Management**: Automatic token storage and refresh for authenticated MCP servers - **Browser-Based Authorization**: Seamless OAuth flow with browser redirects and callback handling - **Server Metadata Support**: Title and metadata fields for better MCP server organization - **Built-in Tool Integration**: Extend AI capabilities with external tools - **Dynamic Server Management**: Add/remove MCP servers through the enhanced UI with connection testing - **Popular MCP Servers**: Support for Composio, Zapier, CogniMemo, and more with improved configuration validation - **Connection Testing**: Test MCP server connections with detailed feedback - **Enhanced UI**: Improved server management interface with validation and OAuth status indicators ### 🎨 User Interface & Experience - **Modern UI**: Built with shadcn/ui components and Tailwind CSS - **Multiple Themes**: Various theme options with enhanced mathematical rendering - **Enhanced Mobile Experience**: Mobile-first responsive design with optimized touch interactions - **Responsive Layout**: Fixed overflow issues on narrow screens with intelligent space management - **Streaming Responses**: Real-time AI response streaming with visual status indicators - **Chat Management**: Persistent chat history and management - **Smart Title Generation**: Dynamic model selection for AI-powered conversation titles - **Enhanced Post-Checkout Navigation**: Better user flow after successful checkout completion - **Enhanced Mathematical Display**: Improved KaTeX styling for consistent mathematical expressions - **Advanced Model Picker**: Search functionality across models with keyboard navigation, provider icons, and improved responsive design with collision handling - **Real-time Streaming Status**: Visual indicators for chat generation progress with timing information - **PDF Download Button**: One-click access to export conversations from chat list ### 📝 Smart Code Input ChatLima automatically detects when you paste code and wraps it in proper markdown code blocks for better formatting: - **Auto-detection**: Automatically detects code based on syntax patterns, keywords, and structure - **Language detection**: Identifies the programming language and adds appropriate syntax highlighting - **Manual wrapping**: Press `Ctrl+K` (or `Cmd+K` on Mac) to manually wrap selected text or entire input in code blocks - **Visual feedback**: Shows a temporary indicator when code is automatically wrapped - **Smart processing**: Normalizes indentation and fixes common copy-paste artifacts #### Supported Languages - JavaScript/TypeScript (including JSX/TSX) - Python - HTML/CSS - JSON - SQL - Shell/Bash - Java - C/C++ - And more... ### 📄 PDF Export & Sharing - **One-Click PDF Download**: Export entire chat conversations as professional PDF documents - **Professional PDF Formatting**: Clean layout with ChatLima.com branding, headers, and page numbers - **Full Markdown Support**: PDFs preserve all markdown formatting including code blocks, headers, and lists - **Automatic Pagination**: Smart page breaks for long conversations - **Social Sharing**: Built-in sharing buttons for Twitter, Facebook, and LinkedIn - **Direct Link Copying**: Copy share URLs with one click for easy distribution - **Client-Side Processing**: All PDF generation happens locally for complete privacy - **Cross-Platform Export**: PDFs work seamlessly across all devices and operating systems ### 🔍 Advanced Capabilities - **Web-Enabled Search**: Enhanced search capabilities via OpenRouter with premium billing - **Native Web Fetch Tool**: Built-in `web_fetch` tool reads and extracts public URLs directly for chat context without MCP dependency - **Reasoning Model Support**: Support for advanced reasoning models with thinking capabilities - **Tool Calling**: MCP server integration for extended capabilities with Grok model support - **Real-time Pricing Analysis**: Developer tools for cost optimization - **Token Usage Tracking**: Accurate billing and usage monitoring - **Smart Credit Exposure**: Frontend access to user credit balance for transparency - **Enhanced Streaming**: Improved handling of interrupted conversations with timeout detection - **Error Recovery**: Intelligent error detection and recovery mechanisms with automatic chat reset - **Advanced Image Analysis**: OCR, visual question answering, and multimodal understanding - **Intelligent Error Recovery**: Comprehensive error handling with automatic recovery mechanisms - **Enhanced UI Resilience**: Error boundary components with graceful error isolation - **Smart Error Detection**: Automatic detection and recovery from stuck or failed requests - **Automatic JSON Repair**: Integrated JSON repair functionality to handle malformed JSON from AI models gracefully - **PDF Export**: One-click PDF generation for chat conversations with professional formatting and markdown support - **Social Sharing**: Built-in sharing buttons for Twitter, Facebook, LinkedIn, and direct link copying ### 🛡️ Security & Privacy - **Secure API Key Management**: Environment-based credential storage with runtime flexibility and show/hide toggles - **Privacy-First Design**: User data protection and anonymous usage support - **Tamper-Proof Daily Limits**: Security patch preventing daily message limit bypass through atomic tracking - **Automated Security Monitoring**: Admin dashboard for user activity monitoring and cleanup management - **SEO Optimization**: Dynamic sitemap generation with privacy protection - **Secure Authentication**: Better Auth with multiple provider support - **Credit Validation**: Robust access control and usage tracking with negative balance prevention - **Environment-Aware Security**: Production/development security configurations - **Image Privacy**: Client-side image processing with secure transmission to AI providers - **API Key Privacy**: Show/hide toggles for sensitive API key management - **Input Validation**: Enhanced validation for all preset and configuration inputs - **Audit Trails**: Comprehensive logging for cleanup operations and usage tracking ### 🚀 Developer Tools - **Real-time Pricing Analysis**: Cost planning and model comparison tools using actual usage data - **Usage Analytics**: Token and cost tracking based on real usage data from 1,254+ API requests - **Debugging Tools**: Enhanced logging and error tracking with comprehensive debugging capabilities - **Development Scripts**: Automated analysis and optimization tools including OpenRouter pricing analysis - **TypeScript Support**: Full TypeScript implementation with comprehensive type definitions - **Testing Infrastructure**: Comprehensive Playwright testing suite with Jest unit tests for services - **Service Test Coverage**: Extensive unit test coverage for all 8 specialized services - **Enhanced Error Boundaries**: Comprehensive error handling with graceful error isolation - **Provider Health Monitoring**: Comprehensive provider status tracking and health checks - **Dynamic Model Architecture**: Extensible provider registry supporting OpenRouter and Requesty with unified interfaces - **Admin CLI Tools**: Command-line utilities for user cleanup and database maintenance - **Health Check APIs**: System monitoring and diagnostics endpoints for operational excellence ### 📊 Business Features - **Dual Subscription Tiers**: Monthly and yearly subscription options for different user segments - **Polar Integration**: Credit purchase and subscription management with production environment - **Usage Monitoring**: Daily/monthly usage tracking and limits with tamper-proof atomic tracking - **Subscription Analytics**: Track subscription types, conversions, and user behavior - **Cost Optimization**: Data-driven model selection and pricing analysis with real usage insights - **Webhook Support**: Real-time payment and subscription updates for both plan types - **Customer Management**: Automated customer creation and management with advanced retrieval logic - **Premium Feature Access**: Intelligent access control for paid features including image processing - **Model Access Control**: Subscription or BYOK gate; per-model credit tiers for usage - **Dynamic Pricing Integration**: Real-time pricing information from provider APIs - **Database Performance Optimization**: Automated user cleanup for optimal database performance and storage efficiency - **Admin Operations Dashboard**: Comprehensive monitoring and management tools for system health ### 🔧 Technical Enhancements - **Modular Service-Oriented Architecture**: Clean, maintainable architecture with 8 specialized services (Authentication, Credit Validation, Database, MCP Server, Message Processing, Model Validation, Token Tracking, Web Search) - **Native URL Extraction Pipeline**: Server-side URL validation, SSRF-safe fetch, readability extraction, and bounded truncation for tool-driven page reading - **Enhanced Error Handling**: Improved error handling across chat API and credit management with domain-specific error classes - **JSON Repair Utility**: Automatic detection and correction of malformed JSON from AI models using jsonrepair library - **Debugging & Traceability**: Enhanced debugging capabilities with comprehensive logging and performant logging utilities - **Smart User Flow Handling**: Seamless experience for both anonymous and authenticated users - **Token Usage Tracking**: Refined credit deduction logic for accurate billing with dedicated tracking service - **Mathematical Content**: Enhanced KaTeX styling for technical discussions - **Progressive Web App**: iOS homescreen shortcut support with native app-like experience - **Advanced Client Instantiation**: Optimized dynamic client creation with helper functions - **Model Transformation Pipeline**: Standardized parsing and transformation of external API responses to internal format - **Intelligent Caching**: Multi-layer caching with 1-hour model list TTL and 24-hour detail caching - **Pre-emptive Operations**: Race condition prevention through pre-emptive chat creation ## Technical Architecture - **Framework**: Next.js 14 with App Router - **AI SDK**: Vercel AI SDK for streaming responses - **Authentication**: Better Auth with Google OAuth - **Database**: Drizzle ORM with PostgreSQL - **Payments**: Polar integration for credits and subscriptions with production environment - **Styling**: Tailwind CSS with shadcn/ui components - **Type Safety**: Full TypeScript implementation - **Testing**: Comprehensive Playwright testing suite with Jest for unit tests - **Image Processing**: Client-side image validation and compression with multimodal AI integration - **Presets System**: Database-backed configuration management with template support - **MCP Integration**: Latest Model Context Protocol 1.13.0 with enhanced server management - **Dynamic Model Loading**: Real-time model fetching with intelligent caching and provider health monitoring - **Service Architecture**: Modular service-oriented design with 8 specialized services for maintainability - **PDF Generation**: jsPDF library for client-side PDF export with markdown rendering - **User Management**: Automated cleanup system with admin dashboard for database optimization - **Error Resilience**: JSON repair utility with jsonrepair library for malformed data handling ## Settings Panel ChatLima features a unified settings panel accessible from the sidebar. Click the "Settings" button in the sidebar to manage all your configuration options in one place. ### Settings Tabs The settings panel includes four main sections: #### API Keys - Manage API keys for 6 AI providers: OpenAI, Anthropic, Groq, XAI, OpenRouter, Requesty - Keys are stored securely in your browser's local storage - Show/hide password toggle for each key - Clear all keys option - Use personal keys to bypass credit system #### MCP Servers - Configure Model Context Protocol servers for extended AI capabilities - Add/edit/delete servers with SSE, stdio, or streamable-http transports - OAuth 2.1 authorization support - Connection testing - Server selection and activation - View active server count badge #### Provider Health - Real-time health monitoring of AI providers - Status indicators: healthy/degraded/down/unknown - Model count per provider - API key source tracking (ENV vs User) - Refresh functionality #### Preferences - Toggle welcome screen visibility - Configure web search context size (Low/Medium/High) ## MCP Server Configuration This application supports connecting to Model Context Protocol (MCP) servers to access their tools. You can add and manage MCP servers through the **Settings → MCP Servers** tab. ### Adding an MCP Server 1. Open Settings from the sidebar 2. Navigate to the "MCP Servers" tab 3. Click "Manage Servers" 4. Enter a name for your MCP server and optionally a display title for better organization. 3. Select the transport type: - **SSE (Server-Sent Events)**: For HTTP-based remote servers - **stdio (Standard I/O)**: For local servers running on the same machine #### SSE Configuration If you select SSE transport: 1. Enter the server URL (e.g., `https://mcp.example.com/token/sse`) 2. **OAuth Authentication** (optional): If the server requires OAuth authentication: - Enable the "Use OAuth Authentication" checkbox - Click "Authorize" to initiate the OAuth flow - You'll be redirected to the server's login page - After successful authentication, tokens will be stored automatically - The server will show an "Authorized" status indicator 3. Click "Add Server" #### stdio Configuration If you select stdio transport: 1. Enter the command to execute (e.g., `npx`) 2. Enter the command arguments (e.g., `-y @modelcontextprotocol/server-google-maps`) - You can enter space-separated arguments or paste a JSON array 3. Click "Add Server" 4. Click "Use" to activate the server for the current chat session. ### OAuth Authorization ChatLima supports OAuth 2.1 authorization for MCP servers that require user authentication. This enables secure access to remote MCP servers like CogniMemo that need user login. #### How OAuth Works 1. **Enable OAuth**: When adding an SSE or Streamable HTTP server, check the "Use OAuth Authentication" option 2. **Authorize**: Click the "Authorize" button to start the OAuth flow 3. **Login**: You'll be redirected to the MCP server's authorization page to log in 4. **Callback**: After successful authentication, you'll be redirected back to ChatLima 5. **Token Storage**: OAuth tokens are securely stored in your browser's localStorage 6. **Automatic Usage**: Tokens are automatically included in all requests to the authenticated server #### OAuth Status Indicators - **Authorized** (green): Server has valid OAuth tokens and is ready to use - **Not authorized** (yellow): OAuth is enabled but no valid tokens exist - click "Authorize" to authenticate #### Token Management - Tokens are automatically refreshed when they expire - Each MCP server's tokens are stored separately and securely - Tokens persist across browser sessions - You can re-authorize at any time if tokens become invalid - **Clear Auth Button**: When a server is authorized, a "Clear Auth" button appears next to the status indicator - Click "Clear Auth" to manually remove all stored OAuth tokens for that specific server - This is useful if you need to switch accounts or reset authentication - After clearing, you'll need to authorize again to use the server ### Enhanced MCP Features - **Connection Testing**: Test MCP server connections with detailed feedback - **Server Metadata**: Add titles and descriptions for better server organization - **Protocol Compliance**: Full MCP 1.13.0 specification support with proper headers - **OAuth Integration**: Seamless OAuth 2.1 flow with automatic token management - **Manual Token Management**: Clear Auth button to manually remove OAuth tokens for individual servers - **Enhanced UI**: Improved server management interface with validation and OAuth status indicators ### Available MCP Servers You can use any MCP-compatible server with this application. Here are some examples: - [Composio](https://composio.dev/mcp) - Provides search, code interpreter, and other tools - [Zapier MCP](https://zapier.com/mcp) - Provides access to Zapier tools - [CogniMemo](https://cognimemo.com) - Memory management and knowledge storage (requires OAuth) - Any MCP server using stdio transport with npx and python3 - Any MCP server using SSE or Streamable HTTP transport (with optional OAuth support) ## License This project is licensed under the Apache License 2.0 - see the [LICENSE](LICENSE) file for details. ## Acknowledgments - Original project by Zaid Mukaddam (https://github.com/zaidmukaddam/scira-mcp-chat) - Built with the [AI SDK by Vercel](https://sdk.vercel.ai/docs) - Powered by [Model Context Protocol (MCP)](https://modelcontextprotocol.io) - UI components from [shadcn/ui](https://ui.shadcn.com/)