Skip to content
Kitbay
Skill

document-extraction

MITUpdated on Nov 10, 202654GitHub starsIncludes scriptsNo API key needed

A skill bundled with fleet-rlm, Qredence's agent runtime, for reading files in the agent's workspace with code instead of loading raw contents.

What the skill teaches

A skill bundled with fleet-rlm, Qredence's agent runtime, for reading files in the agent's workspace with code instead of loading raw contents. A Python helper inspects a file's structure first, such as page, sheet or slide counts and image sizes, then extracts text and tables from PDFs, Word, PowerPoint and Excel files, HTML, Markdown and text. Charts, scans and screenshots go through the runtime's vision path, with large images resized first. The skill says it has no OCR or page rendering, that table extraction needs pdfplumber, and that the Office fallback drops formatting and does not recalculate formulas, and it asks the agent to note which extractor ran and whether output was cut short.

Capabilities

  • Inspect a document's structure before reading it
  • Extract text and tables from PDFs
  • Extract content from Word, PowerPoint and Excel files
  • Inspect and resize workspace images