Feature Guides

Comprehensive guides for all NeuroLink features organized by category. Each guide includes setup, usage patterns, configuration, and troubleshooting.

---

Latest Features (Q1 2026)

Feature	Description
MCP Enhancements	Advanced MCP features: ToolRouter, ToolCache, RequestBatcher, tool annotations, elicitation protocol, and custom MCP server creation.
PPT Generation	Generate professional PowerPoint presentations from text prompts with 35 slide types, 5 themes, and optional AI images.
Video Generation	Generate videos from text prompts using RunwayML (ML5, ML6 Turbo models).
Image Generation with Gemini	Native image generation using Gemini 2.0 Flash Experimental with imagen-3.0-generate-002 model.
HTTP/Streamable HTTP Transport for MCP	Connect to remote MCP servers via HTTP with authentication, rate limiting, retry support, and session management.
Audio Input	Real-time voice conversations with Gemini Live and audio streaming capabilities.
Server Adapters	Expose NeuroLink AI agents as HTTP APIs with Hono, Express, Fastify, and Koa. Production-ready with auth, rate limiting, and streaming.
RAG Document Processing	Comprehensive document chunking (10 strategies), hybrid search (BM25 + vector), and reranking (5 types) for retrieval-augmented generation.
Context Compaction	4-stage context compaction pipeline with automatic budget management, per-provider token estimation, and non-destructive message tagging.
Memory	Per-user condensed memory that persists across conversations. LLM-powered condensation with S3, Redis, or SQLite storage backends.
Claude Subscription Support	Multiple authentication methods for Claude (API key, OAuth) with support for Free, Pro, Max, and API tiers.
Client SDK	Type-safe HTTP, SSE, and WebSocket clients with React hooks and Vercel AI SDK adapter.
Claude Proxy	Multi-account Claude proxy with OAuth pooling, rate-limit failover, token refresh, and launchd daemon for crash recovery.
Authentication Providers	Secure AI endpoints with 11 auth providers (Auth0, Clerk, Firebase, Supabase, Cognito, Keycloak, WorkOS, Better Auth, OAuth2, JWT, Custom) with RBAC, session management, and rate limiting.

Q1 2026 Highlights:

• Authentication Providers: Pluggable multi-provider authentication system with 11 providers, per-call token validation in generate/stream, composable middleware (auth + RBAC + rate limiting), in-memory and Redis session storage, AsyncLocalStorage context propagation, and CLI management commands
• MCP Enhancements: Advanced MCP features including intelligent tool routing (round-robin, least-loaded, capability-based), result caching with LRU/FIFO/LFU eviction, request batching, tool annotations with safety levels, elicitation protocol for interactive input, MCPServerBase for custom servers, and multi-server management with load balancing
• PPT Generation: Create AI-powered PowerPoint presentations with 35 slide types (title, content, charts, timelines, dashboards, composite layouts), 5 built-in themes, optional AI-generated images, and multi-provider support (Vertex, OpenAI, Anthropic, Google AI, Azure, Bedrock)
• Video Generation: Create AI-generated videos with RunwayML integration supporting ML5 and ML6 Turbo models, customizable duration (5-10s), and watermark control
• Gemini Image Generation: Native support for Google's imagen-3.0-generate-002 model through Gemini 2.0 Flash Experimental for high-quality image synthesis
• Remote MCP Servers: HTTP/Streamable HTTP transport enables connecting to cloud-hosted MCP servers with Bearer token authentication, configurable rate limits, automatic retry with exponential backoff, and session management via Mcp-Session-Id header
• Audio Input: Real-time voice conversations with Gemini Live API enabling bidirectional audio streaming for interactive voice-based AI experiences
• Server Adapters: Deploy NeuroLink as production HTTP APIs with support for Hono (recommended), Express, Fastify, and Koa frameworks. Includes built-in authentication, rate limiting, caching, validation middleware, and SSE streaming support.
• RAG Document Processing: Full-featured retrieval-augmented generation with 10 chunking strategies (character, recursive, sentence, token, markdown, html, json, latex, semantic, semantic-markdown), hybrid search combining BM25 and vector similarity, 5 reranking types (simple, LLM, batch, cross-encoder, Cohere), and integration with Pinecone, Weaviate, and Chroma vector stores.
• Claude Subscription Support: Flexible authentication supporting API keys and OAuth for Claude Pro/Max subscribers, with model availability tracking and quota management

---

Core Features (Q4 2025)

Feature	Description
Image Generation	Generate images from text prompts using Gemini models via Vertex AI or Google AI Studio.
Enterprise HITL	Production-ready HITL with approval workflows, confidence thresholds, and enterprise patterns.
Interactive CLI	AI development environment with loop mode, session variables, and conversation memory.
MCP Tools Showcase	Complete guide to 6 built-in tools and 58+ external MCP servers across 6 categories.
Human-in-the-Loop (HITL)	Pause AI tool execution for user approval before risky operations like file deletion or API calls.
Guardrails Middleware	Content filtering, PII detection, and safety checks for AI outputs with zero configuration.
Redis Conversation Export	Export complete session history as JSON for analytics, debugging, and compliance auditing.
Context Compaction	Automatic conversation compression for long-running sessions to stay within token limits.
LiteLLM Integration	Access 100+ AI models from all major providers through unified LiteLLM routing interface.
SageMaker Integration	Deploy and use custom-trained models on AWS SageMaker infrastructure with full control.

---

Core Features (Q3 2025)

Feature	Description
Multimodal Chat Experiences	Stream text and images together with automatic provider fallbacks and format conversion.
CSV File Support	Process CSV files for data analysis with automatic format conversion. Works with all providers.
PDF File Support	Process PDF documents for visual analysis and content extraction. Native provider support.
Office Documents	Process DOCX, PPTX, XLSX files for document analysis. Native Bedrock, Vertex, Anthropic support.
Auto Evaluation Engine	Automated quality scoring and metrics export for AI response validation using LLM-as-judge.
CLI Loop Sessions	Persistent interactive mode with conversation memory and session state for prompt engineering.
Regional Streaming Controls	Region-specific model deployment and routing for compliance and latency optimization.
Provider Orchestration Brain	Adaptive provider and model selection with intelligent fallbacks based on task classification.

---

Platform Capabilities at a Glance

Category	Features	Documentation
Provider unification	14+ providers with automatic failover, cost-aware routing, provider orchestration (Q3)	Provider Setup
Multimodal pipeline	Stream images + CSV data + PDF documents + Office files across providers with auto-detection for mixed file types.	Multimodal Guide, CSV Support, PDF Support, Office Docs
Quality & governance	Auto-evaluation engine (Q3), guardrails middleware (Q4), HITL workflows (Q4), audit logging	Auto Evaluation, Guardrails, HITL
Memory & context	Conversation memory, per-user memory, Mem0 integration, Redis history export (Q4), context summarization (Q4)	Conversation Memory, Memory, Redis Export
CLI tooling	Loop sessions (Q3), setup wizard, config validation, Redis auto-detect, JSON output	CLI Loop, CLI Commands
Enterprise ops	Proxy support, regional routing (Q3), telemetry hooks, configuration management	Enterprise Proxy, Observability
Tool ecosystem	MCP auto discovery, LiteLLM hub access, SageMaker custom deployment, web search	MCP Integration, MCP Catalog

---

AI Provider Integration

NeuroLink supports 13 AI providers with unified API access:

Provider	Key Features	Free Tier	Tool Support	Status	Documentation
OpenAI	GPT-4o, GPT-4o-mini, o1 models	No	Full	Production	Setup Guide
Anthropic	Claude 4.6, 4.5/4.0 Sonnet, Opus, Haiku	No	Full	Production	Setup Guide, Subscription Guide
Google AI	Gemini 2.5 Flash/Pro	Free Tier	Full	Production	Setup Guide
AWS Bedrock	Claude, Titan, Llama, Nova	No	Full	Production	Setup Guide
Google Vertex	Gemini via GCP	No	Full	Production	Setup Guide
Azure OpenAI	GPT-4, GPT-4o, o1	No	Full	Production	Setup Guide
LiteLLM	100+ models unified	Varies	Full	Production	Integration Guide
AWS SageMaker	Custom deployed models	No	Full	Production	Integration Guide
Mistral AI	Mistral Large, Small	Free Tier	Full	Production	Setup Guide
Hugging Face	100,000+ models	Free	Partial	Production	Setup Guide
Ollama	Local models	Free (Local)	Partial	Production	Setup Guide
OpenAI Compatible	Any compatible endpoint	Varies	Full	Production	Setup Guide
OpenRouter	300+ models via unified API	Free Tier	Full	Production	Setup Guide

Provider Comparison Guide - Full feature matrix

---

Advanced CLI Capabilities

Interactive Setup Wizard

NeuroLink includes a revolutionary interactive setup wizard that guides users through provider configuration in 2-3 minutes:

# Launch interactive setup wizard
npx @juspay/neurolink setup

# Provider-specific guided setup
npx @juspay/neurolink setup --provider openai
npx @juspay/neurolink setup --provider bedrock

Wizard Features:

• Secure credential collection with validation
• Real-time authentication testing
• Automatic .env file creation
• Recommended model selection
• Quick-start command examples
• Interactive provider discovery

15+ CLI Commands

Complete command-line toolkit for every workflow:

Command	Description	Key Features
generate/gen	Text generation	Multimodal input, tool support, streaming
stream	Real-time streaming	Live token output, evaluation
loop	Interactive session	Persistent variables, conversation memory
setup	Guided configuration	Provider wizard, validation
status	Health monitoring	Provider health, latency checks
models list	Model discovery	Capability filtering, availability
config	Configuration management	Init, validate, export, reset
memory	Conversation management	Export, import, stats, clear
mcp	MCP server management	List, discover, connect, status
provider	Provider operations	List, test, health dashboard
ollama	Ollama management	Model download, list, remove
sagemaker	SageMaker operations	Status, endpoint management
vertex	Vertex AI operations	Auth status, quota checks
completion	Shell completion	Bash and Zsh support
validate	Config validation	Environment verification

Shell Integration

Bash and Zsh completions for faster command-line workflows:

# Install Bash completion
neurolink completion bash >> ~/.bashrc

# Install Zsh completion
neurolink completion zsh >> ~/.zshrc

Learn more: Complete CLI Reference

---

Built-in Tools & MCP Integration

8 Core Built-in Agent Tools

Complete autonomous agent foundation with security and validation:

Tool	Function	Capabilities	Security	Status
getCurrentTime	Time access	Date/time with timezone support	Safe	Active
readFile	File reading	Secure file system access with path validation	Sandboxed	Active
writeFile	File writing	File creation and modification with safety checks	HITL	Active
listFiles	Directory listing	Directory navigation and listing	Restricted	Active
createDirectory	Directory creation	Directory creation with permission checks	Validated	Active
deleteFile	File deletion	File and directory deletion with confirmation	HITL	Active
executeCommand	Command execution	System command execution with safety limits	HITL	Active
websearchGrounding	Web search	Google Vertex web search integration	API-based	Active

Tool Management System:

• Dynamic tool registration and validation
• Secure execution with sandboxing
• Result processing and error recovery
• Tool discovery and availability tracking

Custom Tools Guide - Create your own tools

---

Model Context Protocol (MCP) - Enterprise-Grade Ecosystem

5 Built-in MCP Servers

NeuroLink includes 5 production-ready MCP servers for enterprise agent deployment:

Server	Purpose	Tools Provided	Status
AI Core	Provider orchestration	generate, select-provider, check-status	Operational
AI Analysis	Analytics capabilities	analyze-usage, performance-metrics	Operational
AI Workflow	Workflow automation	execute-workflow, batch-process	Operational
Direct Tools	Agent integration	file-ops, web-search, execute	Operational
Utilities	General utilities	time, calculations, formatting	Operational

Advanced MCP Infrastructure

Component	Capabilities	Status
Tool Registry	Tool registration, execution, statistics	Active
External Server Manager	Lifecycle management, health monitoring	Active
Tool Discovery Service	Automatic tool discovery and registration	Active
MCP Factory	Lighthouse-compatible server creation	Active
Flexible Tool Validator	Universal safety validation	Active
Context Manager	Rich context with 15+ fields	Active
Tool Orchestrator	Sequential pipelines, error handling	Active

Lighthouse MCP Compatibility

• Factory Pattern: createMCPServer() fully compatible with Lighthouse architecture
• Transport Mechanisms: stdio, HTTP/Streamable HTTP, SSE, WebSocket support (99% compatibility)
• Tool Standards: Full MCP specification compliance
• Context Passing: Rich context with sessionId, userId, permissions (15+ fields)

58+ External MCP Servers

Supported for extended functionality:

Categories:

• Development: GitHub, GitLab, filesystem access
• Databases: PostgreSQL, MySQL, SQLite
• Cloud Storage: Google Drive, AWS S3
• Communication: Slack, email
• And many more...

Quick Example:

// Add any MCP server dynamically
await neurolink.addExternalMCPServer("github", {
  command: "npx",
  args: ["-y", "@modelcontextprotocol/server-github"],
  transport: "stdio",
  env: { GITHUB_TOKEN: process.env.GITHUB_TOKEN },
});

// Tools automatically available to AI
const result = await neurolink.generate({
  input: { text: 'Create a GitHub issue titled "Bug in auth flow"' },
});

MCP Integration Guide - Setup and usage MCP Server Catalog - Complete server list (58+)

---

Developer Experience Features

SDK Features

Feature	Description	Documentation
Auto Provider Selection	Intelligent provider fallback	SDK Guide
Streaming Responses	Real-time token streaming	Streaming Guide
Conversation Memory	Automatic context management	Memory Guide
Full Type Safety	Complete TypeScript types	Type Reference
Error Handling	Graceful provider fallback	Error Guide
Analytics & Evaluation	Usage tracking, quality scores	Analytics Guide
Middleware System	Request/response hooks	Middleware Guide
Framework Integration	Next.js, SvelteKit, Express	Framework Guides

---

CLI Features

Feature	Description	Documentation
Interactive Setup	Guided provider configuration	Setup Guide
Text Generation	CLI-based generation	Generate Command
Streaming	Real-time streaming output	Stream Command
Loop Sessions	Persistent interactive mode	Loop Sessions
Provider Management	Health checks and status	CLI Guide
Model Evaluation	Automated testing	Eval Command
MCP Management	Server discovery and installation	MCP CLI

15+ Commands for every workflow - see Complete CLI Reference

---

Smart Model Selection & Cost Optimization

Cost Optimization Features

• Automatic Cost Optimization: Selects cheapest models for simple tasks
• LiteLLM Model Routing: Access 100+ models with automatic load balancing
• Capability-Based Selection: Find models with specific features (vision, function calling)
• Intelligent Fallback: Seamless switching when providers fail

CLI Examples:

# Cost optimization - automatically use cheapest model
npx @juspay/neurolink generate "Hello" --optimize-cost

# LiteLLM specific model selection
npx @juspay/neurolink generate "Complex analysis" --provider litellm --model "anthropic/claude-3-5-sonnet"

# Auto-select best available provider
npx @juspay/neurolink generate "Write code" # Automatically chooses optimal provider

Learn more: Provider Orchestration Guide

---

Interactive Loop Mode

NeuroLink features a powerful interactive loop mode that transforms the CLI into a persistent, stateful session.

Key Capabilities

• Run any CLI command without restarting session
• Persistent session variables: set provider openai, set temperature 0.9
• Conversation memory: AI remembers previous turns within session
• Redis auto-detection: Automatically connects if REDIS_URL is set
• Export session history as JSON for analytics

Quick Start

# Start loop with Redis-backed conversation memory
npx @juspay/neurolink loop --enable-conversation-memory --auto-redis

# Start loop without Redis auto-detection
npx @juspay/neurolink loop --enable-conversation-memory --no-auto-redis

Example Session

# Start the interactive session
$ npx @juspay/neurolink loop

neurolink » set provider google-ai
✓ provider set to google-ai

neurolink » set temperature 0.8
✓ temperature set to 0.8

neurolink » generate "Tell me a fun fact about space"
The quietest place on Earth is an anechoic chamber at Microsoft's headquarters...

# Exit the session
neurolink » exit

Complete Loop Guide - Full documentation with all commands

---

Enterprise & Production Features

Production Capabilities

Feature	Description	Use Case	Documentation
Enterprise Proxy	Corporate proxy support	Behind firewalls	Proxy Setup
Redis Memory	Distributed conversation state	Multi-instance deployment	Redis Guide
Cost Optimization	Automatic cheapest model selection	Budget control	Cost Guide
Multi-Provider Failover	Automatic provider switching	High availability	Failover Guide
Telemetry & Monitoring	OpenTelemetry integration	Observability	Observability Guide
Security Hardening	Credential management, auditing	Compliance	Security Guide
Custom Model Hosting	SageMaker integration	Private models	SageMaker Guide
Load Balancing	LiteLLM proxy integration	Scale & routing	Load Balancing Guide
Audit Trails	Comprehensive logging	Compliance	Audit Guide
Configuration Management	Environment & credential management	Multi-environment deployment	Config Guide

Advanced Security Features

Human-in-the-Loop (HITL) Policy Engine

Enterprise-grade approval system for sensitive operations:

// HITL Policy Configuration
type HITLPolicy = {
  requireApprovalFor: string[]; // Tool-specific policies
  autoApprove: string[]; // Safe operation whitelist
  alwaysDeny: string[]; // Blacklist operations
  timeoutBehavior: "deny" | "approve"; // Timeout handling
};

HITL Capabilities:

• User consent for dangerous operations
• Configurable policy engine
• Comprehensive audit trail logging
• Timeout handling
• Bulk approval for batch operations

Advanced Proxy Support

Corporate network compatibility:

Proxy Type	Support	Features
AWS Proxy	Full	AWS-specific proxy configuration
HTTP/HTTPS Proxy	Full	Universal proxy across all providers
No-Proxy Bypass	Full	Bypass configuration and utilities

Enhanced Guardrails

AI-powered content security:

• Content Filtering: Automatic content screening
• Toxicity Detection: Toxic content filtering
• PII Redaction: Privacy protection and PII detection
• Custom Rules: Configurable policy rules
• Security Reporting: Detailed security event reporting

Security & Compliance Certifications

• SOC2 Type II compliant deployments
• ISO 27001 certified infrastructure compatible
• GDPR-compliant data handling (EU providers available)
• HIPAA compatible (with proper configuration)
• Hardened OS verified (SELinux, AppArmor)
• Zero credential logging
• Encrypted configuration storage

Enterprise Deployment Guide - Complete production patterns

---

Middleware & Extension System

Advanced Middleware Architecture

Pluggable request/response processing for custom workflows:

Built-in Middleware

Middleware	Purpose	Features	Status
Analytics	Usage tracking & monitoring	Token counting, timing, performance metrics	Active
Guardrails	Content security	Content policies, toxicity detection, PII filtering	Active
Auto Evaluation	Quality scoring	LLM-as-judge, accuracy metrics, safety validation	Active

Middleware System Capabilities

// Middleware Configuration
type MiddlewareFactoryOptions = {
  middleware?: NeuroLinkMiddleware[]; // Custom middleware registration
  enabledMiddleware?: string[]; // Selective activation
  disabledMiddleware?: string[]; // Selective deactivation
  middlewareConfig?: Record<string, MiddlewareConfig>; // Per-middleware configuration
  preset?: string; // Preset configurations
  global?: {
    // Global settings
    maxExecutionTime?: number;
    continueOnError?: boolean;
  };
};

Middleware Features:

• Dynamic middleware registration
• Pipeline execution with performance tracking
• Runtime configuration changes
• Error handling and graceful recovery
• Priority-based execution order
• Detailed execution statistics

Custom Middleware Guide - Build your own middleware

---

Performance & Optimization

Intelligent Cost Optimization

• Model Resolver: Cost optimization algorithms and intelligent routing
• Performance Routing: Speed-optimized provider selection
• Concurrent Initialization: Reduced latency through parallel loading
• Caching Strategies: Intelligent response and configuration caching

Advanced SageMaker Features

Beyond basic integration - enterprise-grade custom model deployment:

Feature	Description	Status
Adaptive Semaphore	Dynamic concurrency control for optimal throughput	Implemented
Structured Output Parser	Complex response parsing and validation	Implemented
Capability Detection	Automatic endpoint capability discovery	Implemented
Batch Inference	Efficient batch processing for high-volume workloads	Implemented
Diagnostics System	Real-time endpoint monitoring and debugging	Implemented

Error Handling & Resilience

Production-grade fault tolerance:

• MCP Circuit Breaker: Fault tolerance with state management
• Error Hierarchies: Comprehensive error types for HITL, providers, and MCP
• Graceful Degradation: Intelligent fallback strategies
• Retry Logic: Configurable retry with exponential backoff

Performance Optimization Guide - Complete optimization strategies

---

Advanced Integrations

Integration	Description
LiteLLM Integration	Access 100+ models from all major providers via LiteLLM routing with unified interface.
SageMaker Integration	Deploy and call custom endpoints directly from NeuroLink CLI/SDK with full control.
Memory	Per-user condensed memory with S3/Redis/SQLite storage and LLM-powered condensation.
Enterprise Proxy	Configure outbound policies and compliance posture for corporate environments.
Configuration Management	Manage environments, regions, and credentials safely across deployments.

---

Advanced Features

Feature	Description
🏭 Factory Pattern Architecture	Unified provider interface with automatic fallbacks and type-safe implementations.
🗄️ Conversation Memory	Deep dive into memory management and Redis integration.
🔧 Custom Middleware	Build request/response hooks for logging, filtering, and custom processing.
⚡ Performance Optimization	Caching, connection pooling, and latency optimization strategies.
📊 Telemetry & Observability	OpenTelemetry integration for distributed tracing and monitoring.
🧪 Testing Guide	Provider-agnostic testing, mocking, and quality assurance strategies.
📊 Analytics & Evaluation	Usage tracking, cost monitoring, and quality scoring for AI responses.
⚡ Streaming	Real-time token streaming with provider-specific optimizations.
Thinking Configuration	Configure extended thinking levels for supported models (Anthropic, Gemini 2.5+).
Structured Output	JSON schema-based structured output with provider-specific formatting.
Text-to-Speech (TTS)	Basic TTS support via Google Cloud TTS (Neural2, Wavenet, Standard voices).

---

See Also

• Getting Started - Quick start and installation
• CLI Reference - Command-line interface documentation
• SDK Reference - TypeScript API documentation
• Enterprise Guides - Production deployment patterns
• Tutorials - Step-by-step implementation guides
• Examples - Real-world code samples