filex

Parse documents, media files, and URLs into Markdown with the FileX CLI.

1.2k|126|Updated Mar 14, 2025
One-click install
npx skills add https://github.com/inclusionAI/AWorld --skill filex
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: filex
Source: https://github.com/inclusionAI/AWorld/tree/main/aworld-skills/filex
Command: npx skills add https://github.com/inclusionAI/AWorld --skill filex

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Converting heterogeneous files—PDFs, Office documents, spreadsheets, images, audio, video, and YouTube links—into readable Markdown normally requires juggling many different parsers and OCR tools. This Skill wraps the FileX CLI so an agent can parse, inspect, and track all of these sources through one validated interface inside an AWorld sandbox.

Core Features & Use Cases

  • Universal file parsing: Convert PDF, DOCX, PPTX, XLSX, CSV, images, audio, and video files from the workspace or HTTP(S) URLs into Markdown, with selectable providers such as Paddle OCR, LiteParse, and local Whisper.
  • YouTube transcript extraction: Inspect video metadata and subtitle tracks, then parse the best available transcript into timestamped Markdown without downloading media unless a rights basis is confirmed.
  • Resumable PDF batch processing: Split large PDFs into page batches, run them asynchronously, and poll progress with a stable batch resume id.
  • Use Case: A researcher drops a 200-page scanned PDF report into the workspace and asks the agent to extract its contents; the Skill parses it in batches with OCR and returns a Markdown file the agent can summarize.

Quick Start

Use the filex skill to parse /root/workspace/report.pdf into Markdown and summarize its key findings.

Frequently Asked Questions about filex

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a PDF or Word document to Markdown in a sandbox?

Run the filex.py wrapper with the parse command, passing the workspace file path via --input and an optional --output destination. FileX detects the file type automatically and returns JSON containing the output_path of the generated Markdown.

How to extract a YouTube transcript without downloading the video?

Use the inspect command with the video URL to view subtitle tracks, then run parse with --mode transcript and a preferred --language. FileX prefers human subtitles over automatic captions and only downloads media when you explicitly confirm a rights basis.

What file formats does FileX support for parsing?

FileX supports PDF, TXT, Markdown, DOC/DOCX, PPT/PPTX, CSV, XLS/XLSX, common image formats (PNG, JPEG, WebP, GIF, BMP), audio (MP3, WAV, M4A, FLAC, OGG), and video (MP4, MOV, MKV, WebM, AVI).

Can I process a large PDF in batches and resume later?

Yes, pass --page-batch-size and a stable --batch-resume-id with --sync-mode async to parse in background batches. Poll progress anytime with the status command using the same resume id.

Why does the filex parse command fail with a dependency error?

The wrapper requires the FileX executable in the sandbox image; if command -v filex finds nothing, parsing cannot proceed. Use a FileX-enabled sandbox image rather than falling back to unsupported built-in parsers.