Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/manavsehgal/strands-analyst

Analyst AI agentic system powered by Strands Agents
https://github.com/manavsehgal/strands-analyst

Last synced: 12 days ago
JSON representation

Analyst AI agentic system powered by Strands Agents

Awesome Lists containing this project

README 

          
# Strands Analyst






![Python 3.8+](https://img.shields.io/badge/python-3.8%2B-blue)

![Version](https://img.shields.io/badge/version-0.1.0--alpha-blue)

![License](https://img.shields.io/badge/license-MIT-green)

![AWS Bedrock](https://img.shields.io/badge/AWS-Bedrock-orange)

![Anthropic API](https://img.shields.io/badge/Anthropic-API-blue)

![OpenAI API](https://img.shields.io/badge/OpenAI-API-green)

![Strands Agents](https://img.shields.io/badge/powered%20by-Strands%20Agents-purple)



**Next-Generation Multi-Provider GenAI & Agentic AI Toolkit**



*Built with performance, security, and enterprise scalability in mind*



[🚀 Quick Start](#-quick-start) • [🎯 Features](#-features) • [📊 Architecture](#-architecture) • [🔧 Installation](#-installation) • [📖 Documentation](#-documentation) • [🛠️ CLI Tools](#️-cli-tools)






---



## 🎯 Overview



**Strands Analyst** is a cutting-edge AI platform designed for AWS Solutions Architects, GenAI professionals, and enterprise teams building scalable AI solutions. Built on the powerful **Strands Agents framework** with **multi-provider model support**, it provides specialized CLI tools, an interactive AI assistant with 40+ professional-grade tools, and production-ready configurations optimized for enterprise GenAI workflows.



### 🌟 What Makes Strands Analyst Unique?



- **🌐 Multi-Provider AI Models**: Seamlessly switch between AWS Bedrock, Anthropic API, and OpenAI API

- **🧰 44+ AI Tools**: Comprehensive toolkit spanning RAG & memory, multimodal AI, automation, and system integration

- **🎨 Smart Organization**: AI-generated files automatically categorized into type-specific directories

- **⚡ Dynamic Intelligence**: Automatic model selection based on task complexity analysis

- **🔧 Enhanced Feedback**: Rich, colored tool execution display with error explanations

- **☁️ Enterprise-Ready**: Production configurations with advanced caching, streaming, and cost optimization



## ✨ Latest Features



### 🌐 Multi-Provider Model Support

**Choose the best AI model for your needs**:

- 🚀 **AWS Bedrock** - Enterprise-grade with guardrails, caching, and Claude 3.7 Sonnet

- 🤖 **Anthropic API** - Direct Claude access with Sonnet 4 and structured output

- 🌟 **OpenAI API** - GPT-4o with function calling and o1-preview reasoning models

- 🔄 **Dynamic switching** via environment variables or configuration

- 📊 **Health monitoring** and cost optimization across providers



```bash

# Switch providers instantly

STRANDS_PROVIDER=openai analystai "Generate a Python script"

STRANDS_PROVIDER=anthropic analystai "Analyze this data"

STRANDS_PROVIDER=bedrock analystai "Extract metadata"



# Check provider health

provider-info --health-check

```



### 📁 Smart File Organization

**AI-generated files are automatically organized**:

- 🎯 **50+ file types** recognized and categorized automatically

- 📁 **Type-based directories** - Code, markdown, data, images, diagrams

- 📅 **Optional date organization** for time-series data

- 🔍 **Respects explicit paths** when user specifies them



```

analystai-responses/

├── code/           # Python, JavaScript, etc.

├── markdown/       # Documentation, notes

├── data/           # JSON, YAML, XML files

├── images/         # Generated visualizations

└── diagrams/       # Architecture diagrams

```



### 🔧 Enhanced Tool Output Display

**Rich feedback for every operation**:

- 🎨 **Colored terminal output** with tool identification

- 📝 **Input visualization** - URLs (🌐), files (📄), text (📝)

- ❌ **Smart error messages** with explanations (404, timeout, DNS)

- ⏱️ **Optional timing** for performance monitoring



```bash

🔧 Tool: fetch_url_metadata

   🌐 Url: https://anthropic.com

   ✅ Metadata extracted successfully



🔧 Tool: save_file_smart

   📄 File: analystai-responses/code/analysis.py

   📝 Text: # Data analysis script...

```



## 🚀 Quick Start



### Installation & Setup



```bash

# Create and activate virtual environment

python -m venv .venv

source .venv/bin/activate  # Windows: .venv\Scripts\activate



# Install with all features

pip install -e .



# Configure your preferred AI provider

export OPENAI_API_KEY="sk-..."        # For OpenAI

export ANTHROPIC_API_KEY="sk-ant-..."  # For Anthropic

aws configure                          # For Bedrock

```



### Interactive AI Assistant



```bash

# Start the enhanced AI assistant

analystai



# Try these example prompts from 200+ curated examples:

> "Draw me an enterprise RAG architecture using Bedrock Knowledge Bases"

> "Compare Claude vs GPT-4o costs for 1 million users monthly"  

> "Generate fractals and save them as images"

> "Create a GenAI transformation roadmap presentation"

> "Analyze anthropic.com and stripe.com websites"

> "Take a screenshot using shell automation"

```



### Specialized CLI Tools



```bash

# Website Intelligence & Analysis

sitemeta anthropic.com --verbose

sitemeta stripe.com --save-markdown



# Multi-Provider Content Analysis  

STRANDS_PROVIDER=openai news https://feeds.bbci.co.uk/news/rss.xml

STRANDS_PROVIDER=anthropic article https://aws.amazon.com/blogs/machine-learning/



# HTML Processing & Conversion

htmlmd saved-article/index.html --no-metadata



# Check provider configuration

provider-info --verbose --health-check

```



## 🎯 Features



### 🤖 Interactive AI Assistant (`analystai`)



The crown jewel of Strands Analyst - an advanced conversational AI with **44+ specialized tools** across **10 categories**, featuring **multi-provider model support** and **intelligent tool execution feedback**.



#### 🎯 **Comprehensive Use Case Examples**



Built-in **200+ curated prompts** for real-world GenAI workflows:



🏗️ GenAI Architecture & Design



- "Draw me an enterprise RAG architecture using Bedrock Knowledge Bases and Claude"

- "What would a conversational AI platform look like on AWS with Bedrock and API Gateway?"

- "Show me how to design a multi-modal GenAI system that handles text, images, and video"

- "Create a 3-tier scalable GenAI application architecture with auto-scaling"



🤖 Agentic Architecture & Automation



- "How can I automate our customer support workflows using Bedrock Agents?"

- "I need multiple AI agents working together to generate and review content automatically"  

- "Design an intelligent document processing system that can take actions based on what it reads"

- "Create a network of specialized AI agents for our content creation pipeline"



💰 Cost Analysis & Optimization



- "Compare Bedrock Claude vs Titan costs for an enterprise chatbot serving 1 million users monthly"

- "Calculate the total cost and ROI of deploying Amazon Q Business for our 5000-person company"

- "Model how GenAI infrastructure costs scale as we grow from startup to enterprise scale"

- "Compare multi-provider costs: OpenAI vs Anthropic vs Bedrock for our workload"



🔒 Security & Compliance



- "Create a GenAI governance framework for healthcare with Bedrock Guardrails"

- "How do I protect PII data when using Bedrock Knowledge Bases in my enterprise system?"

- "Build algorithms to detect prompt injection attacks in our GenAI applications"

- "Generate a compliance checklist for responsible AI implementation"



#### 📦 **Complete Tool Categories**



🧠 RAG & Memory Systems



- `retrieve` - Semantic search and retrieval from knowledge bases

- `memory` - Session-based memory management  

- `agent_core_memory` - Persistent agent memory across sessions

- `mem0_memory` - Advanced memory storage with contextual understanding



📁 File Operations



- `file_read` - Secure file reading with permission controls

- `file_write` - Safe file writing with consent management

- `save_file_smart` - Automatic file organization by type

- `editor` - Interactive file editing capabilities



⚙️ System & Automation



- `shell` - Execute shell commands with security consent

- `use_computer` - Computer automation and control

- `cron` - Task scheduling and automation

- `environment` - Environment variable management



🌐 Web & Network



- `http_request` - HTTP/API requests and integrations

- `browser` - Web browsing and page interaction

- `rss` - RSS feed monitoring and analysis

- `slack` - Slack integration and notifications



🎨 Multimodal Capabilities



- `diagram` - Generate professional architecture diagrams

- `generate_image` - AI-powered image generation (fractals, visualizations)

- `speak` - Text-to-speech conversion

- `image_reader` - Image analysis and OCR

- `nova_reels` - Video content generation



#### ✨ Enhanced Chat Experience

- 🎨 **Rich Terminal UI** with beautiful panels and color-coded output

- ⚡ **Real-time streaming** responses as they generate

- 🔧 **Live tool indicators** showing active operations in progress

- 📝 **Markdown rendering** for beautifully formatted content

- 🌐 **Provider display** showing which AI model is active

- 🔄 **Stable fallback modes** ensuring compatibility



### 🛠️ CLI Tools



Professional command-line tools for specialized workflows:



#### 🌐 `sitemeta` - Website Intelligence

```bash

sitemeta google.com                    # Basic site analysis

sitemeta stripe.com --verbose          # Detailed analysis with metrics

sitemeta anthropic.com --save-markdown # Save results to markdown

```

*Analyze websites to understand business models, extract metadata, and generate intelligence reports.*



#### 📰 `news` - RSS & News Analysis  

```bash

news https://feeds.bbci.co.uk/news/rss.xml                    # Analyze RSS feed

news https://aws.amazon.com/blogs/ml/feed/ --count 10         # Latest 10 articles

news https://example.com/feed --save-markdown --verbose       # Full analysis with save

```

*Fetch, analyze, and summarize RSS feeds and news sources with AI-powered insights.*



#### 📄 `article` - Web Article Processing

```bash

article https://example.com/blog-post                         # Download and analyze

article https://aws.amazon.com/blogs/ml/post --no-images     # Skip image downloads  

article https://medium.com/@author/post --verbose             # Detailed processing info

```

*Download web articles with metadata extraction, image preservation, and content analysis.*



#### 📝 `htmlmd` - HTML to Markdown Conversion

```bash

htmlmd saved-article/index.html                               # Convert to markdown

htmlmd document.html --no-metadata                            # Skip metadata extraction

htmlmd content.html --output custom-output.md --verbose       # Custom output with details

```

*Convert HTML files to clean, well-formatted markdown with metadata preservation.*



#### 🔧 `provider-info` - Multi-Provider Management

```bash

provider-info                          # Show active provider

provider-info --verbose                # Detailed model information  

provider-info --health-check           # Test provider connectivity

STRANDS_PROVIDER=openai provider-info  # Check specific provider

```

*Monitor and manage multi-provider AI model configurations.*



#### 📄 `pdf-to-markdown` - PDF Document Processing

```bash

pdf-to-markdown document.pdf                     # Convert PDF to markdown

pdf-to-markdown report.pdf --verbose             # Show processing details

pdf-to-markdown paper.pdf --save-markdown        # Save to organized directory

```

*Convert PDF documents to clean, structured markdown with intelligent text extraction.*



## 🏗️ Architecture & Performance



### 🌐 Multi-Provider Architecture



```mermaid

graph TD

    A[Strands Analyst] --> B[Provider Factory]

    B --> C[AWS Bedrock]

    B --> D[Anthropic API]

    B --> E[OpenAI API]

    

    C --> F[Claude 3.7 Sonnet]

    C --> G[Claude 3.5 Haiku]

    

    D --> H[Claude Sonnet 4] 

    D --> I[Claude Opus 4.1]

    

    E --> J[GPT-4o]

    E --> K[GPT-4o-mini]

```



**Provider-Specific Features**:



| Feature | AWS Bedrock | Anthropic API | OpenAI API |

|---------|-------------|---------------|-------------|

| **Models** | Claude 3.7 Sonnet, 3.5 Haiku | Sonnet 4, Opus 4.1, Haiku | GPT-4o, GPT-4o-mini, o1-preview |

| **Streaming** | ✅ | ✅ | ✅ |

| **Function Calling** | ✅ | ❌ | ✅ |

| **Structured Output** | ✅ | ✅ | ✅ |

| **Guardrails** | ✅ | ❌ | ❌ |

| **Caching** | ✅ | ❌ | ❌ |

| **Tool Use** | ✅ | ❌ | ✅ |

| **Cost** | Enterprise | Pay-per-use | Pay-per-use |



### ⚡ Performance Optimizations



#### Dynamic Model Selection

- **Task complexity analysis** automatically selects optimal models

- **Model warm-up capabilities** eliminate cold start latency  

- **Runtime configuration updates** without application restart

- **Agent-specific tuning**: Temperature, top_p, and token limits optimized per use case



#### Smart File Organization  

- **Automatic categorization** of 50+ file types into appropriate directories

- **Type-based routing**: Code → `code/`, Data → `data/`, Images → `images/`

- **Date-based organization** optional for time-series data

- **User path preservation** when explicit paths are specified



#### Enhanced Tool Feedback

- **Real-time tool identification** with colored terminal output

- **Input categorization** with icons: URLs (🌐), files (📄), text (📝)

- **Intelligent error explanations** for common failures (404, timeout, DNS)

- **Performance timing** optional for bottleneck identification



#### Advanced Caching & Optimization

- **Multi-level caching**: System prompts, tool definitions, and message-level caching

- **Streaming responses** for improved perceived performance

- **Concurrent tool execution** for multi-tool workflows

- **Intelligent context management** reducing token usage by up to 40%



### 📊 Enterprise Observability



- **Real-time metrics tracking** with token consumption and tool performance analytics

- **Cost monitoring** and budget alerts across all providers

- **Performance regression detection** with automated optimization recommendations  

- **Multi-provider health checks** ensuring system reliability



### 🏗️ Modular Architecture



```

strands-analyst/

├── analyst/

│   ├── agents/              # AI agent implementations

│   │   ├── chat.py         # Interactive AI assistant with multi-provider

│   │   ├── sitemeta.py     # Website analysis agent

│   │   ├── news.py         # RSS/news analysis agent

│   │   └── get_article.py  # Article processing agent

│   ├── tools/              # Reusable tool implementations  

│   │   ├── fetch_url_metadata.py    # Efficient metadata extraction

│   │   ├── save_file_smart.py       # Smart file organization

│   │   └── download_article_content.py # Content downloading

│   ├── utils/              # Core utilities

│   │   ├── model_provider_factory.py  # Multi-provider management

│   │   ├── smart_file_saver.py        # File organization logic

│   │   └── tool_output_display.py     # Enhanced tool feedback

│   └── cli/                # Command-line interfaces

│       ├── chat.py         # analystai command with provider switching

│       └── provider_info.py # provider-info command

├── docs/                   # Comprehensive documentation

├── analystai-responses/    # Auto-organized output files

│   ├── code/              # Python, JavaScript, etc.

│   ├── markdown/          # Documentation, analyses

│   ├── data/              # JSON, YAML, XML files

│   ├── images/            # Generated visualizations

│   └── diagrams/          # Architecture diagrams

└── refer/                 # Sample outputs and examples

    ├── articles/          # Downloaded web articles (70+ examples)

    ├── sitemeta/          # Website analyses

    └── posts/             # Generated content and studies

```



## 🎨 Generated Content Examples



### 📊 **Visualizations & Fractals**



The system can generate beautiful mathematical visualizations and fractals:



![Complex Julia Fractal](analystai-responses/images/complex_julia_fractal.png)

*Complex Julia fractal generated with AI-powered mathematical visualization*



![Mandelbrot Deep Sea](analystai-responses/images/mandelbrot_zoom_deep_sea.png)  

*Deep zoom Mandelbrot set with custom color mapping*



### 📈 **Business Analysis Charts**



Professional charts and visualizations for business intelligence:



![NVIDIA Meta RoCE Components](analystai-responses/images/nvidia_meta_roce_components.png)

*Technology component analysis with automated chart generation*



### 📄 **Generated Content Samples**



Real examples from the `analystai-responses/markdown/` directory:



- **[AWS MoE LLM Architecture](analystai-responses/markdown/aws_moe_llm_architecture.md)** - Comprehensive Mixture-of-Experts implementation guide

- **[Character.AI Platform Overview](analystai-responses/markdown/character-ai-overview.md)** - AI platform competitive analysis  

- **[NVIDIA Meta RoCE Analysis](analystai-responses/markdown/nvidia_meta_roce_analysis.md)** - Technical deep-dive with visualizations

- **[Apple Models Analysis](analystai-responses/markdown/apple-models.md)** - Apple's AI model ecosystem analysis

- **[AWS Architecture for Agentic AI](analystai-responses/markdown/aws-architecture-agentic-ai-application.md)** - Enterprise-grade agentic AI system design



### 🏢 **Real-World Article Processing**



The `refer/articles/` directory contains **70+ processed articles** demonstrating comprehensive content extraction:



- **Enterprise case studies**: [HubSpot's Story](refer/articles/about-hubspot-hubspots-story/), [Decagon Conversational AI](refer/articles/about-decagon-conversational-ai-for-cx/)

- **Technical deep-dives**: [AI Workflow Automation](refer/articles/ai-workflow-automation-platform-tools-n8n/), LLM architectures

- **Industry analyses**: [AI Business Models](refer/articles/ai-business-informs-educates-and-connects-the-global-ai-comm/), technology trends



**Sample Processing Results**:

- **Metadata extraction**: Title, description, author, publish date

- **Content preservation**: Full article text with formatting

- **Image downloads**: Local copies of all article images

- **Structured output**: Clean markdown with organized file structure



## 🔧 Installation



### Prerequisites



- **Python 3.8+** (recommended: Python 3.11 or 3.13)

- **AI Provider Access**:

  - AWS Account with Bedrock access (Claude 3.7 Sonnet enabled)

  - Anthropic API key (optional)

  - OpenAI API key (optional)

- **System Dependencies**: See advanced features section



### Quick Installation



```bash

# Create and activate virtual environment (recommended)

python -m venv .venv

source .venv/bin/activate  # On Windows: .venv\Scripts\activate



# Install Strands Analyst with all community tools

pip install -e .



# Verify installation

analystai --help

provider-info --health-check

```



### 🔑 API Key Configuration



#### Multi-Provider Setup

```bash

# Option 1: Environment variables (recommended)

export OPENAI_API_KEY="sk-proj-..."

export ANTHROPIC_API_KEY="sk-ant-..."

# AWS credentials via aws configure



# Option 2: .env.local file

echo 'OPENAI_API_KEY=sk-proj-...' >> .env.local

echo 'ANTHROPIC_API_KEY=sk-ant-...' >> .env.local



# Option 3: Configuration file (not recommended for security)

# Edit config.yml with API keys

```



#### Provider Selection

```bash

# Set default provider in config.yml

providers:

  active: "openai"  # or "anthropic" or "bedrock"



# Or override with environment variable

export STRANDS_PROVIDER=anthropic

```



### 📦 Advanced Dependencies



#### For Diagram Generation (Required for `diagram` tool)

```bash  

# macOS

brew install graphviz



# Ubuntu/Debian  

sudo apt-get install graphviz



# Windows

# Download from: https://graphviz.org/download/



# Verify installation

dot -V

```



#### For Browser Automation (Required for `browser` and `use_computer` tools)

```bash

# Install Playwright browsers for web automation

playwright install



# Verify browser installation

playwright list

```



#### For PDF Processing (Required for `pdf_to_markdown` tool)

```bash

# Included with installation via pymupdf4llm

# Verify with:

python -c "import pymupdf4llm; print('PDF processing ready')"

```



#### AWS Configuration

```bash

# Configure AWS credentials (for Bedrock)

aws configure



# Test Bedrock access

aws bedrock list-foundation-models --region us-west-2



# Verify Claude 3.7 Sonnet access

aws bedrock get-foundation-model --model-identifier anthropic.claude-3-7-sonnet-20250219-v1:0

```



## 📖 Documentation



### 📚 Core Documentation

- **[Installation Guide](docs/installation.md)** - Complete setup instructions

- **[CLI Guide](docs/cli-guide.md)** - Command-line interface usage

- **[Configuration Guide](docs/configuration-guide.md)** - Advanced configuration options

- **[Developer Guide](docs/developer-guide.md)** - Extending with new agents and tools



### ✨ New Features

- **[Multi-Provider Model Guide](docs/multi-provider-guide.md)** - Switch between AWS Bedrock, Anthropic, and OpenAI providers 🚀

- **[Smart File Organization Guide](docs/file-organization-guide.md)** - Automatic file categorization and directory management 📁

- **[Enhanced Tool Output Guide](docs/tool-output-guide.md)** - Rich, colored feedback for tool execution 🔧



### 🔧 Enhanced Features

- **[Enhanced Chat Features Guide](docs/enhanced-chat-guide.md)** - Rich terminal UI with streaming support and 40+ tools

- **[Community Tools Guide](docs/community-tools-guide.md)** - Complete 44+ tools integration

- **[Automation Guide](docs/automation-guide.md)** - Computer & browser automation via shell



### 🎯 Agent-Specific Guides

- **[Chat Agent Guide](docs/chat-agent-guide.md)** - Interactive conversational interface

- **[Article Agent Guide](docs/article-agent-guide.md)** - Web article processing  

- **[HTML to Markdown Guide](docs/htmlmd-agent-guide.md)** - HTML conversion features

- **[News Agent Guide](docs/news-agent-guide.md)** - RSS feed analysis

- **[PDF to Markdown Guide](docs/pdf-to-markdown-guide.md)** - PDF document processing and conversion



## 🚀 Use Cases



### For AWS Solutions Architects

- **Multi-Provider Strategy**: Compare Bedrock vs Anthropic vs OpenAI for specific workloads

- **Architecture Planning**: Generate AWS GenAI architecture diagrams with provider-specific components

- **Cost Optimization**: Analyze costs across providers for enterprise deployments

- **Technology Research**: Stay updated with latest AI/ML services across cloud providers



### For GenAI Professionals  

- **Model Experimentation**: Test the same prompt across multiple providers instantly

- **Content Intelligence**: Analyze websites and articles for competitive research

- **Performance Optimization**: Leverage dynamic model selection and advanced caching

- **Multi-modal Workflows**: Combine text, image, and diagram generation seamlessly



### For Enterprise Teams

- **Unified AI Interface**: Single CLI for all major AI providers

- **Organized Output**: Automatic file categorization keeps projects clean  

- **Advanced Monitoring**: Provider health checks and performance tracking

- **Security Compliance**: Secure tool execution with consent management



## 🛡️ Security & Compliance



- **🔐 Multi-Provider Security**: Separate credential management for each provider

- **🔒 Secure Tool Execution**: Sandboxed environment for code execution

- **🛡️ Consent Management**: User approval required for system-level operations  

- **📊 Audit Logging**: Comprehensive logging across all providers

- **⚠️ Provider Guardrails**: Bedrock guardrails integration when available



## 🗺️ Roadmap



### 🚧 Recently Completed ✅

- **Multi-provider model support** with OpenAI, Anthropic, and Bedrock  

- **Smart file organization** with automatic type-based categorization

- **Enhanced tool output display** with rich terminal feedback

- **Dynamic model configuration** with complexity-based selection

- **Provider health monitoring** and switching capabilities



### 🔄 In Development

- **Message-level caching** for conversation continuity

- **Multi-agent orchestration** framework with agents-as-tools pattern

- **OpenTelemetry integration** for standardized instrumentation  

- **Real-time cost tracking** across all providers



### 🔮 Future Plans

- **Mixture-of-Experts** architecture support across providers

- **Edge computing patterns** for sub-100ms response times

- **Advanced memory systems** with long-term context retention

- **Custom model fine-tuning** workflows



## 🤝 Contributing



We welcome contributions! Areas of interest:

- New AI provider integrations

- Additional tool categories  

- Performance optimizations

- Documentation improvements



Please see our contribution guidelines and feel free to submit issues and pull requests.



## 📄 License



This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.



## 🆘 Support



- **Documentation**: Comprehensive guides available in `/docs`

- **Provider Issues**: Use `provider-info --health-check` for diagnostics

- **GitHub Issues**: Report bugs and request features

- **Community**: Join our discussions for tips and best practices



---






**Built with ❤️ for the future of enterprise AI**



*Strands Analyst - Where Multi-Provider AI meets Performance Excellence*