axoviq-ai/synthadoc

DOCX extracts paragraph text from .docx files using python-docx. Use it when you need plain text from a Microsoft Word document or when the input path ends with .docx.

Image extracts text and key information from images with a vision LLM. Use it for screenshots, diagrams, photos, or any .png, .jpg, .webp, .gif, or .tiff input.

Markdown reads Markdown or plain text files and returns their contents verbatim. Use it when a source path ends in .md or .txt, or when you need to ingest notes or text files.

PDF extracts text from PDF files using pypdf, with pdfminer.six as a fallback for CJK fonts. Use it when the source is a .pdf file or the user asks for a research paper.

Pptx extracts text from .pptx presentations using python-pptx. Use it to turn slide titles, body text, and speaker notes into plain text.

Session extracts human-readable conversation turns from AI coding session history files in JSONL format. Use it to convert Claude Code, Codex, or Cursor session logs into cleaned user and assistant text.

URL fetches web pages and extracts clean body text from HTML or PDF URLs. Use it when a source is a web address and you need readable text for downstream processing.

Web Search accepts a natural-language query, calls Tavily, and returns matching URLs. Use it when you need candidate sources to fetch or pass to another skill.

Xlsx extracts rows from .xlsx and .csv files into comma-separated text. Use it when you need to read spreadsheet data from Excel or CSV inputs.

YouTube extracts captions from YouTube video URLs without an API key or audio download. Use it to ingest a video's transcript, and optionally generate a structured summary with an LLM.