Community Research MCP

# ============================================================================ # REQUIRED: At least ONE LLM API Key # ============================================================================ # Option 1: Gemini (Recommended - FREE) # - 1,500 requests/day free # - No credit card required # - Get key: https://makersuite.google.com/app/apikey GEMINI_API_KEY= # Option 2: OpenAI (Paid - Best Quality) # - Costs money per request # - Get key: https://platform.openai.com/api-keys OPENAI_API_KEY= # Option 3: Anthropic Claude (Paid - High Quality) # - Costs money per request # - Get key: https://console.anthropic.com/ ANTHROPIC_API_KEY= # Option 4: OpenRouter (Free $5 credit, then paid) # - Access to 100+ models # - Get key: https://openrouter.ai/keys OPENROUTER_API_KEY= # Option 5: Perplexity (5 requests/day free, then paid) # - Has web search built-in # - Get key: https://www.perplexity.ai/settings/api PERPLEXITY_API_KEY= # ============================================================================ # OPTIONAL: Enhanced Search APIs (Dramatically Improves Results!) # ============================================================================ # Serper (Google Search API - HIGHLY RECOMMENDED ⭐) # - 2,500 free searches per month # - Access to Google Search results # - Dramatically better results than free APIs # - Get key: https://serper.dev/ SERPER_API_KEY= # Brave Search (Privacy-focused alternative) # - 2,000 free searches per month # - Privacy-focused search engine # - Get key: https://brave.com/search/api/ BRAVE_SEARCH_API_KEY= # ============================================================================ # OPTIONAL: Reddit API Access (Enhanced Reddit Functionality) # ============================================================================ # Reddit API OAuth Credentials # - Required for authenticated Reddit access # - Higher rate limits and full comment tree access # - Get credentials: https://www.reddit.com/prefs/apps REDDIT_CLIENT_ID= REDDIT_CLIENT_SECRET= REDDIT_REFRESH_TOKEN= # ============================================================================ # OPTIONAL: Azure OpenAI (Enterprise/Corporate Users) # ============================================================================ # Azure OpenAI Service # - Enterprise-grade OpenAI models # - Private deployment in your Azure subscription # - Enhanced security and compliance # - Get from Azure Portal: https://portal.azure.com/ AZURE_OPENAI_ENDPOINT= AZURE_OPENAI_API_KEY= AZURE_OPENAI_API_VERSION=2024-02-15-preview AZURE_OPENAI_DEPLOYMENT_NAME= # ============================================================================ # OPTIONAL: Ollama (Local AI Models - Privacy & Cost Savings) # ============================================================================ # Ollama Local Server # - Run AI models locally on your computer # - Complete privacy (no data sent to external APIs) # - Zero ongoing costs after initial setup # - Install: https://ollama.ai/ OLLAMA_ENDPOINT=http://localhost:11434 OLLAMA_MODEL=llama2 # ============================================================================ # MULTI-MODEL ORCHESTRATION SETTINGS # ============================================================================ # Model Selection Strategy # - auto: Automatically select best model based on query complexity # - manual: Use single provider specified in priority # - consensus: Get opinions from multiple models MODEL_SELECTION_STRATEGY=auto # Provider Priority (comma-separated, first = highest priority) # - Controls which provider to try first when auto-selecting # - Available: gemini,openai,anthropic,azure,ollama,openrouter,perplexity PROVIDER_PRIORITY=gemini,openai,anthropic,azure,ollama # Thinking Mode (affects analysis depth vs speed/cost) # - quick: Fast responses, lower cost, basic analysis # - balanced: Default mode, good balance of depth and speed # - deep: Thorough analysis, higher cost, maximum insight DEFAULT_THINKING_MODE=balanced # ============================================================================ # OPTIONAL: Multi-Model Validation (DISABLED BY DEFAULT - Costs 2x API calls) # ============================================================================ # When enabled, a second AI model critiques the first model's findings # Use for: Security-critical code, authentication flows, high-stakes decisions # Cost: Doubles API usage (runs 2 models instead of 1) ENABLE_MULTI_MODEL_VALIDATION=false VALIDATION_PROVIDER=gemini # ============================================================================ # NOTES ON IMPROVING RESULTS # ============================================================================ # # By default, the server uses these FREE APIs: # - Stack Overflow (limited technical coverage) # - GitHub Issues/Discussions (limited scope) # - Reddit (hit or miss) # - Hacker News (limited coverage) # # To get MUCH BETTER results: # 1. Add SERPER_API_KEY (gives you Google Search access) # 2. Add PERPLEXITY_API_KEY (has built-in web search) # 3. Use broader search terms for niche topics # 4. Enable MULTI_MODEL_VALIDATION for critical research # # With SERPER_API_KEY + Multi-Model, you'll get 10x more results! #