Mace-legal/mike

Adds redline-aware document extraction for DOCX tracked changes and PDF color annotations

Active work on document ingestion. The fork focuses on a specific gap in the upstream codebase: reviewed documents (contracts under negotiation, annotated drafts) arrive with edit markers that the base parser discards. This fork keeps them visible to the model.

View on GitHub →

Mace-legal/mike extends the base extraction pipeline to surface redlines directly in the assistant context. DOCX tracked-change markers and PDF color-annotated segments are both parsed and inserted as inline markers rather than silently stripped. The primary goal is giving the assistant visibility into both what has changed in a document and its current state.

What's in it

Tracked changes awareness Word documents come in with their tracked edits intact, so the AI can reason about what's being proposed, not just the final text.
Margin comments in scope Reviewer comments attached to a Word doc are surfaced alongside the body, giving the assistant the full negotiation thread.
Colour-coded PDF redlines PDFs marked up with coloured redlines are parsed as redlines - not as flat text with mysterious highlights.

Direction

contract-reviewdrafting

Activity

Themed changes and pull requests touching this fork, newest first. Themed changes that haven't been turned into a public post yet still appear — they're real work even without a published writeup.

📝 Mace-legal teaches Mike to read redlines 1 commit 2mo ago contract-review draft

The fork now understands tracked changes and reviewer comments instead of flattening them into noise.

✅ #1 feat: add redline-aware document extraction 2mo ago self by JonasBoury ↗ analysis ↗ GitHub

## Summary

📝 Document extraction now preserves tracked changes and comments instead of flattening them 0 commits — contract-reviewdiscovery draft

JonasBoury landed a change that stops the ingest pipeline from silently discarding review markup. DOCX tracked changes and comment bubbles now come through as readable inline markers, and PDFs with color-coded annotation…

Threads of work (detailed view)

3 threads have been distilled into posts.

Mace-legal teaches Mike to read redlines

The fork now understands tracked changes and reviewer comments instead of flattening them into noise.

contract-review

1 commit · latest 2mo ago

Redline-aware extraction: DOCX tracked changes and PDF color annotations surfaced as inline markers

JonasBoury added end-to-end support for preserving review markup through the document ingest pipeline. DOCX tracked changes and reviewer comments now arrive at the LLM as inline markers rather than getting flattened; PDFs with color-based redlines (Litera/Workshare style) get a PyMuPDF extractor, with the existing pdfjs path kept as fallback.

contract-reviewdiscovery

1 commit · latest 2mo ago

Document extraction now preserves tracked changes and comments instead of flattening them

contract-reviewdiscovery

Pull requests (detailed view)

1 PR touch this fork — inbound (filed against it) or outbound (filed from it). State icons match the editorial dashboard.

✅ Merged (1)

✅ merged · #1 feat: add redline-aware document extraction

by JonasBoury · opened 2mo ago · merged 2mo ago

Mace-legal/mike ← Mace-legal/mike · self · ↗ analysis · ↗ GitHub