Invoice Extractor

Extract structured data from invoices and receipts (PDFs and images). Output JSON, CSV, or build a running expense ledger. Use when someone shares an invoice...

This skill appears coherent and implements the described features locally. Before installing/running: (1) review that you are comfortable with the ledger being written on disk (default: skill_dir/data/ledger.csv) or pass a --config to change the path; (2) be aware that image processing uses the agent's vision tool — images may be sent to the platform/provider for analysis; (3) batch mode processes files and asks a single confirmation at the end (no per-file prompts) — test with a small folder first; (4) install pdfplumber or PyPDF2 only if you trust running third-party Python packages; (5) future versions mention OAuth and email integration — grant those credentials only if you understand and trust the skill's export flows. If you want additional assurance, run the script in a restricted environment (or review the remainder of the script beyond the provided truncation) before giving it access to your data.

Type: OpenClaw Skill Name: rebels-invoice-extractor Version: 1.2.0 The 'rebels-invoice-extractor' skill is a well-structured tool designed for extracting data from invoices and managing a local expense ledger. The core logic in 'scripts/extract.py' uses standard libraries (pdfplumber, PyPDF2, csv, json) to process files and maintain a CSV-based ledger with features like duplicate detection and platform-specific exports (Xero, FreeAgent, etc.). The 'SKILL.md' instructions follow a safe workflow, explicitly requiring the agent to obtain user confirmation before committing any extracted data to the ledger. No evidence of data exfiltration, malicious code execution, or unauthorized persistence was found; the 'discovery flow' for new export formats is a legitimate use of the agent's capabilities to update its own configuration.