PDFCompress v0.1.0

PDFCompress v0.1.0

PDFCompress is an open-source, structure-preserving PDF compressor designed for both people and AI
agents. This first release provides a conservative compression pipeline that keeps searchable text,
vectors, annotations, bookmarks, attachments, fonts, and page geometry intact while recompressing
only eligible oversized images.

Highlights

• WPS-compatible image presets: high, standard, medium, and low.
• An additional archive-safe preset for formal, award, tender, and archival attachments.
• Optional target-size search that always starts from the original input.
• Verification of page count, page boxes, and extractable text for every candidate.
• Explicit protection against accidental source overwrite and signature invalidation.
• CLI commands for analyze, compress, and verify.
• Agent integration through an MCP server, shared skill, Codex plugin, and Claude Code plugin.
• Cross-platform CI on Windows, Linux, and macOS with Python 3.11 and 3.13.

Install

uv tool install "pdfcompress-agent[mcp] @ git+https://github.com/ncepuee/PDFCompress.git@v0.1.0"

Verify downloads

The release includes a source distribution, a Python wheel, and SHA256SUMS-v0.1.0.txt. Verify an
asset on Windows with:

Get-FileHash .\pdfcompress_agent-0.1.0-py3-none-any.whl -Algorithm SHA256

Notes

• PDF rewriting invalidates digital signatures; signed inputs are rejected unless explicitly allowed.
• Version 0.1 conservatively skips unusual image objects rather than risking visual corruption.
• Pixel-level visual review remains recommended for legally significant or highly complex documents.