feat: langchain integration

[ritankarsaha]

📌 Description

Please include a summary of the changes and the related issue. Provide enough context for reviewers to understand what the PR is doing.

• What problem does it solve?

Fixes #[issue-number] (if applicable)

---

✅ Changes

Complete LangChain Integration Structure

📁 Directory Structure

  lib/langchain/
  ├── config.ts              # Centralized configuration
  ├── types.ts               # Comprehensive type definitions
  ├── models/
  │   └── index.ts           # Model factory (OpenAI/Anthropic)
  ├── agents/
  │   └── base-agent.ts      # Powerful agent implementation
  ├── retrievers/
  │   ├── vector-store.ts    # Multi-vector store support
  │   └── knowledge-base.ts  # Advanced knowledge base retriever
  ├── tools/
  │   ├── index.ts           # Tool registry and factory
  │   ├── fact-check.ts      # Comprehensive fact-checking
  │   ├── search.ts          # Advanced search capabilities
  │   └── analytics.ts       # Data analysis tools
  ├── utils/
  │   ├── logger.ts          # Advanced logging system
  │   ├── error-handler.ts   # Comprehensive error handling
  │   └── index.ts           # Utility collections
  └── index.ts               # Main export and quick helpers

  services/
  └── agents.ts              # High-level agent service orchestration

Key Features Implemented

1. Powerful Agent Setup

• Multi-provider support: OpenAI & Anthropic Claude
• 3 specialized agents: General, Researcher, Analyst
• Advanced memory management: Buffer & Summary memory
• Streaming support: Real-time response generation
• Tool integration: Automatic tool discovery and usage

2. Knowledge Base Retriever

• Multi-vector store support: Pinecone, Chroma, Qdrant, Memory
• Advanced search: Filtering, ranking, faceted search
• Document preprocessing: Chunking, overlap handling
• Re-ranking capabilities: Embeddings-based re-ranking
• Performance optimized: Caching and pagination

3. Comprehensive Fact-Checking Tools

• Multi-source verification: Web search, Knowledge base, Manual
  surces
• Source credibility scoring: Automated trust assessment
• Evidence analysis: AI-powered verdict determination
• Confidence scoring: Transparent reliability metrics
• Structured output: JSON schema with detailed metadata

4. Advanced Error Handling & Logging

• Circuit breaker pattern: Prevents cascading failures
• Retry with exponential backoff: Resilient error recovery
• Comprehensive logging: Structured logs with performance tracking
• Error pattern analysis: Automatic issue detection
• Health monitoring: Real-time system status

5. Production-Ready Features

• Environment validation: Config verification on startup
• Performance monitoring: Execution time tracking
• Security utilities: Input sanitization and validation
• Modular architecture: Easy to extend and maintain
• TypeScript throughout: Full type safety

Usage Examples

Quick Setup

import { initializeLangChain, quickChat, quickFactCheck } from
'@/lib/langchain';

// Initialize everything
const { agentService, isHealthy } = await initializeLangChain();

// Quick chat
const response = await quickChat("What is quantum computing?",
"researcher");

// Quick fact-check
const factCheck = await quickFactCheck("The Earth is flat");

Advanced Usage

import { AgentService, FactCheckTool, KnowledgeBaseRetriever } from
'@/lib/langchain';

// Custom agent service
const agentService = new AgentService();
await agentService.initialize();

// Specialized fact-checking
const factCheckResult = await agentService.factCheck({
claim: "AI will replace all jobs by 2030",
context: "Economic analysis context",
options: {
maxSources: 10,
confidenceThreshold: 0.9
}
});

// Advanced search
const searchResults = await agentService.search({
query: "machine learning algorithms",
filters: {
categories: ["research", "technical"],
dateRange: { start: new Date("2023-01-01"), end: new Date() }
}
});

🛠️ Configuration

The system supports extensive configuration through environment
variables:

Model Configuration

OPENAI_API_KEY=your_openai_key
ANTHROPIC_API_KEY=your_anthropic_key
PREFER_OPENAI=false

Vector Store Configuration

VECTOR_STORE_PROVIDER=pinecone
PINECONE_API_KEY=your_pinecone_key
VECTOR_INDEX_NAME=knowledge-base

Search Configuration

SERPAPI_API_KEY=your_serpapi_key
GOOGLE_SEARCH_ENGINE_ID=your_search_engine_id

Logging Configuration

LOG_LEVEL=info
ENABLE_AGENT_LOGS=true
LOGGING_WEBHOOK_URL=your_webhook_url

The integration is production-ready, fully modular, and extensively
documented. It provides powerful AI agents with fact-checking
capabilities, knowledge base integration, and enterprise-grade error
handling and monitoring.

Checklist

• I have tested this code
• I have added necessary documentation
• I have linked relevant issue(s)
• I followed the code style of the project
• I reviewed my own code