An AI-powered document processing assistant with specialized skills for working with Office files, PDFs, and data visualization.
Select from various AI models in the UI:
- OpenAI: GPT-5.1, GPT-5.2, GPT-5.3
- Anthropic: Claude Sonnet 4
- Google: Gemini 3.1 Pro
- Word (.docx): Create, edit, format documents
- PowerPoint (.pptx): Create and modify presentations
- PDF: Read, merge, split, fill forms
- Excel (.xlsx): Data analysis, formulas, formatting
- Matplotlib: Static charts (PNG) at 300 DPI for professional quality
- Plotly: Interactive visualizations (HTML)
- Embedded in Excel: Insert high-quality charts directly into spreadsheets
- LibreOffice integration for formula recalculation
- File upload support (documents and images)
- Download generated files from sidebar
- Persistent outputs directory
- Docker deployment ready
- Install dependencies:
pip install -e .- Set up API keys (create
.envfile):
ANTHROPIC_API_KEY=your_key
OPENAI_API_KEY=your_key
GOOGLE_API_KEY=your_key- Run the app:
streamlit run app.pySee DOCKER_GUIDE.md for complete Docker instructions.
docker-compose up --build- Select Model: Choose your preferred AI model from the sidebar
- Upload Files: Optional - upload documents or images to process
- Chat: Ask the agent to create, edit, or analyze documents
- Download: Get generated files from the sidebar or chat
Create a sales report in Excel with:
- Q1-Q4 sales data
- Bar chart showing revenue by quarter
- Line chart for trend analysis
Save to outputs/sales_report.xlsx
Create a professional business proposal in Word with:
- Cover page with title "Digital Transformation Proposal"
- Table of contents
- 3 sections: Executive Summary, Approach, Timeline
- Save as outputs/proposal.docx
[Upload a PDF file]
Summarize the key points from this document and create a Word summary
.
├── app.py # Streamlit UI
├── main.py # Agent configuration
├── agent_skills/ # Specialized skills
│ ├── docx/ # Word document skills
│ ├── pptx/ # PowerPoint skills
│ ├── pdf/ # PDF processing skills
│ └── xlsx/ # Excel skills
├── outputs/ # Generated files (auto-created)
└── Dockerfile # Docker configuration
Each skill provides specialized capabilities:
- docx: Word document manipulation with python-docx
- pptx: PowerPoint creation and editing with python-pptx
- pdf: PDF processing with pypdf and pdfplumber
- xlsx: Excel analysis and creation with pandas/openpyxl
- validators: Shared utilities for Office document validation
- Python 3.12+
- LibreOffice (for Excel formula recalculation)
- API keys for at least one model provider
- Create a directory in
agent_skills/ - Add a
SKILL.mdfile with instructions - Include any reference scripts in
scripts/ - The agent will automatically load the skill
Run the app locally and test with various document types:
streamlit run app.pySee individual skill directories for licenses (most use MIT or Apache 2.0).
For issues or questions, check the skill documentation in agent_skills/*/SKILL.md.