Release v1.0.6

Release v1.0.6

Highlights

A new read_file copyable tool that reads any file type and returns raw extracted content straight to your agent — text, code, Markdown, CSV, JSON, LaTeX source, images, PDF, PowerPoint, Word, audio, and video — all through one co copy read_file.

What's Changed

✨ Features

• read_file multi-format tool (#170) — one tool that dispatches by extension:

  • Text / code / .md / .csv / .json / .tex → returned as-is, with encoding auto-detected (UTF-8 or GB18030 for Chinese — no mojibake)
  • Images (.png / .jpg / .gif / .webp) → data:image/...;base64 URL for the vision model
  • PDF → per-page text (pypdf)
  • PowerPoint (.pptx) → slide text (python-pptx)
  • Word (.docx) → document text (python-docx)
  • Audio → transcript via transcribe()
  • Video → ffmpeg pulls the audio track, then transcribe()

  Content is returned raw with no truncation. Registered in the co copy tools registry (not re-exported from useful_tools, so it doesn't collide with the built-in text read_file).

📚 Documentation

• Document co email name and co email upgrade
• Fix co email upgrade docs — --domain is required for plus/pro

Installation

pip install connectonion==1.0.6

Breaking Changes

None.

Full Changelog: v1.0.5...v1.0.6