Split out from the #404 bug-audit re-review (see docs/BUG_AUDIT.md → I1).
Status: single-source skip_scrape is now honored in SkillConverter.run() (PR #404). The unified multi-source path (UnifiedScraper) still ignores it and re-scrapes every source from the network.
Why deferred: honoring it requires reloading each source's .skillseeker-cache into scraped_data before build_skill() — a real feature, not a one-liner, with its own test surface.
Fix sketch: in UnifiedScraper.run(), when self.skip_scrape, skip scrape_all_sources() and instead load each source's cached data from .skillseeker-cache/.../sources/, then build. Guard against missing cache.
Split out from the #404 bug-audit re-review (see
docs/BUG_AUDIT.md→ I1).Status: single-source
skip_scrapeis now honored inSkillConverter.run()(PR #404). The unified multi-source path (UnifiedScraper) still ignores it and re-scrapes every source from the network.Why deferred: honoring it requires reloading each source's
.skillseeker-cacheintoscraped_databeforebuild_skill()— a real feature, not a one-liner, with its own test surface.Fix sketch: in
UnifiedScraper.run(), whenself.skip_scrape, skipscrape_all_sources()and instead load each source's cached data from.skillseeker-cache/.../sources/, then build. Guard against missing cache.