This release adds 3 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+9 more
ReleasePort's take
Light signaldocling v2.95.0 preserves text in DrawingML DOCX files and fixes CLI version detection when meta-package is missing. Release adds scoped settings context manager, expanded metadata in callbacks, and custom headers support for HTML image fetching.
Why it matters: DOCX users with DrawingML images should test this release to verify text extraction. Callback-based workflows must review metadata expansion impact. CLI users should upgrade to resolve version detection failures.
Summary
AI summaryPreserve text in DOCX files when image uses DrawingML without LibreOffice.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Medium |
Adds scoped settings context manager. Adds scoped settings context manager. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Feature | Medium |
Adds more document metadata in callbacks. Adds more document metadata in callbacks. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Feature | Medium |
Adds custom headers support for HTML backend image fetching. Adds custom headers support for HTML backend image fetching. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Bugfix | Medium |
Preserves text when image uses DrawingML without LibreOffice in docx output. Preserves text when image uses DrawingML without LibreOffice in docx output. Source: llm_adapter@2026-05-21 Confidence: high |
— |
| Bugfix | Medium |
Fixes unknown docling version issue when meta-package is missing in CLI. Fixes unknown docling version issue when meta-package is missing in CLI. Source: llm_adapter@2026-05-21 Confidence: high |
— |
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
Beta — feedback welcome: [email protected]