AI-workbench [Tomb-stoned]

A reusable library of AI agents for various automation tasks. This workbench includes stateless and stateful agents that can be used for code generation, text summarization, and multi-step task execution.

For Documentation Click Here

Features

• Stateless Agent: Lightweight, one-shot request/response agent for quick queries and code generation
• Summarizer Agent: Advanced planning and execution agent with async support for complex tasks
• Virtual Assistant Agent: Persona-based chatbot that learns over time and maintains memory via a critic-led cycle
• Multi-Provider Support: Works with Gemini, OpenAI, Anthropic, LiteLLM, Ollama, Perplexity, HuggingFace, and more
• Extensible Architecture: Easy to add new agents and tools

Installation

1. Clone the repository:

   git clone <repository-url>
   cd AI-workbench

2. Create and activate a virtual environment:

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

3. Install dependencies:

   uv sync
   # To install optional test dependencies:
   uv sync --all-extras

4. Configure environment variables: Create a .env file with your API keys:

   GOOGLE_API_KEY=your_api_key_here

Usage

Stateless Agent

Stateless Agent

Run the stateless agent for quick, one-off requests:

uv run stateless-agent

Summarizer Agent

Run the summarizer agent for planning and multi-step tasks:

uv run summarizer-agent

Virtual Assistant Agent

Run the virtual assistant agent for persona-based interactions:

uv run virtual-assistant
# or to launch the chatbot interface:
uv run virtual-assistant-chatbot

Documentation

This project uses MkDocs with the Dracula theme for documentation.

View Documentation Locally

mkdocs serve

Then open http://127.0.0.1:8000 in your browser.

Build Documentation

mkdocs build

Deploy to GitHub Pages

mkdocs gh-deploy

Quick MkDocs Notes

• Configuration: mkdocs.yml
• Documentation source: docs/ folder
• User guides: docs/user-guide/
• API reference: docs/api-reference/
• Current theme: Dracula (dark mode)

Project Structure

AI-workbench/
├── src/
│   ├── agents/          # Agent implementations
│   ├── apps/            # Application interfaces runner
│   ├── core/            # Core modules (providers, tools, utils)
│   ├── config/          # Configuration settings
│   ├── interfaces/      # Base interfaces
│   └── run_*.py         # Agents Runner scripts
├── docs/                # MkDocs documentation
├── tests/               # Unit tests
└── requirements.txt     # Python dependencies

License

See LICENSE file for details.