Repository navigation
Add document adapters and evaluate the SourcePack workflow - #1
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
SourcePack previously turned a long transcript into hundreds of tiny inspection steps, and a blocked video download could make useful captions inaccessible. This change groups timed evidence, keeps partial acquisitions usable, and adds optional local document conversion and portable evidence packs.
The skill now includes bounded scene/overview/targeted frames, simpler read/record packets, explicit supporting sources, and Agent Skills-style frozen evaluations. MarkItDown's pinned qualification build passed DOCX, PPTX, XLSX and text-PDF fixtures. Docling remains disabled pending offline model qualification; OCR is not advertised as working.
Validation: 84 default tests passed, two optional live tests passed when enabled, skill format validation passed, and the final copied skill completed real PPTX prepare/read/record/answer/export outside the repository. A fresh reviewer identified five issues, all reproduced and repaired with regressions.
The paired pilot found equal selected-answer quality. SourcePack took 447s versus 121s for the held-out video and both configurations recovered a visual setting after restart. The results explicitly make no accuracy, speed, cost or adoption advantage claim. Private source media and evaluation traces are excluded from Git.
See docs/acceptance/2026-09-19-results.md for measurements, caveats and the recommendation to keep the skill small and reduce interaction overhead.