Our review
Extracts structured financial data from receipt and invoice images using OCR or AI vision.
Strengths
- Automated extraction of vendor name, date, total, tax, and line items
- Local OCR mode to save API tokens
- Built-in risk assessment
- Flexible configuration via environment variables or CLI arguments
Limitations
- Requires an API key for vision mode (unless using local OCR)
- Output quality depends on image clarity
- Only supports JPG, PNG, and WebP formats
To quickly digitize and categorize expenses from receipt photos.
If images are blurry or low-resolution, or if you need a fully offline solution without any LLM API.
Security analysis
CautionThe skill runs a Python script to analyze receipt images, which involves file access and potential external API calls. This poses a moderate risk of sensitive data exposure, but no destructive actions are instructed.
- •Executes Python script with user-provided file path
- •May upload receipt images to external AI service using API keys
- •Environment variables could expose API credentials
Examples
Analyze the receipt at C:/Users/Docs/receipt_123.jpg and tell me the total amount.Extract all item names and prices from the invoice image /home/user/invoice.png.Assess the risk of the receipt at receipts/dinner_2025.jpg and explain any anomalies.TaxHacker Receipt Analysis Skill
Extract structured financial data from receipt/invoice images using AI.
Description
This skill analyzes images of receipts, invoices, or bills to extract key accounting information including vendor name, date, total amount, currency, tax, and itemized lists. It also provides an AI-based risk assessment and categorization.
Configuration (OpenClaw)
This skill is optimized for OpenClaw. If you have configured a model in OpenClaw, the skill can automatically use its settings.
- If
OPENCLAW_LLM_API_KEYis set in the environment, the skill will use it. - If
OPENCLAW_LLM_BASE_URLis set, the skill will use it. - If
OPENCLAW_LLM_MODELis set, the skill will use it (e.g.,gpt-4o,gpt-4o-mini).
Otherwise, you can provide parameters directly via CLI:
# Vision Mode (Default)
python -m tax_hacker_skill.skill "<file_path>" --api-key "YOUR_KEY" --model "gpt-4o-mini"
# Local OCR Mode (Saves tokens, no image upload)
python -m tax_hacker_skill.skill "<file_path>" --local-ocr
# Local OCR Mode with CnOCR
python -m tax_hacker_skill.skill "<file_path>" --local-ocr --ocr-engine cnocr
Parameters
file_path: (Required) Absolute path to the receipt image file (jpg, png, webp).--api-key: (Optional) Override API Key.--base-url: (Optional) Override API Base URL.--model: (Optional) Override LLM Model.--local-ocr: (Optional) Use local OCR instead of vision.--ocr-engine: (Optional) OCR engine:auto,easyocr,cnocr.
Instructions
- Ensure the user has provided an image file path.
- Execute the following command to process the receipt:
python -m tax_hacker_skill.skill "<file_path>" - The command will output a JSON object containing the extracted data.
- Use the output to update accounting records, generate reports, or answer user queries about their expenses.
Example Usage
"Analyze the receipt at C:/Users/Docs/receipt_123.jpg and tell me the total amount."
-> Run: python -m tax_hacker_skill.skill "C:/Users/Docs/receipt_123.jpg"
Prompt Engineering
Data & AI
Prompt engineering best practices and templates to maximize AI outputs.
Data Visualization
Data & AI
Generates data visualizations and charts tailored to your data.
RAG Architecture Setup
Data & AI
Setup guide for RAG (Retrieval-Augmented Generation) architectures.