Home
A tool for tracking and analyzing differences across documentation versions using semantic content extraction.
Features¶
- Hash-based delta detection: Quickly identify changed, added, removed, and renamed documents
- Semantic content comparison: Extract and compare content blocks (headings, text, code, tables) while ignoring cosmetic HTML changes
- Fuzzy rename detection: Identify renamed/moved documents using similarity matching
- Automated polling daemon: Monitor GraphQL APIs for documentation changes and trigger pipelines automatically
- Structured reporting: Generate detailed JSON reports with change analysis
- QA generation (optional): Generate question-answer pairs from documentation changes using RAGAS
Quick Start¶
One-command pipeline¶
uv run docta diff full \
--old-root data/docs_v1 \
--new-root data/docs_v2 \
--old-version "1.0" \
--new-version "2.0" \
--output-dir artifacts \
--max-docs 50 \
--allow-overwrite
This creates two reports:
artifacts/delta_report.json— hash-based change detection (what changed)artifacts/semantic_diff_report.json— detailed semantic analysis (how it changed)
Two-stage pipeline¶
# Stage 1: Generate delta report
uv run docta diff compare \
--old-root data/docs_v1 \
--new-root data/docs_v2 \
--old-version "1.0" \
--new-version "2.0" \
--output artifacts/delta_report.json \
--allow-overwrite
# Stage 2: Semantic content comparison
uv run docta diff scan \
--report artifacts/delta_report.json \
--old-root data/docs_v1 \
--new-root data/docs_v2 \
--output artifacts/semantic_diff_report.json \
--max-docs 50 \
--allow-overwrite
Next steps¶
- Installation — requirements and setup
- Document Comparison — how semantic diffing works and all options
- GraphQL Daemon — automated polling and pipeline configuration
- QA Generation — generating QA pairs from documentation changes
- Commands Reference — full command listing