ShadowCrawl v2.4.3
v2.4.3 (2026-02-19)
Feat
- GitHub blob URL auto-rewrite:
web_fetchongithub.com/*/blob/*URLs is now transparently rewritten toraw.githubusercontent.combefore fetching — returns the raw file/source directly instead of GitHub's React SPA shell. - GitHub SPA payload extraction:
looks_like_spanow detects GitHub'sreact-app.embeddedDatascript tag.extract_spa_json_stateextractspayload.blob.text,payload.readme,payload.issue.body,payload.pullRequest.body,payload.discussion.bodyfrom the embedded JSON — gives clean readable content on repo home pages, issues, and PR pages. - JSON fragment line filter in
post_clean_text: Lines that are ≥55% JSON structural tokens ({}"[]:,) and lines starting with{or[with length >40 are stripped as pipeline noise — prevents escaped GitHub SPA fragments from leaking intoclean_content.
Chore (build hygiene)
-
web_fetch:extract_relevant_sections=truereturns only the most relevant sections forquery(short output; avoids huge tool responses). -
Fixes cross-target build warnings caused by platform-specific
cfgblocks:- removes
unused_importsforPath/PathBufon non-macOS targets - avoids
dead_codewarnings for setup-only helpers on Windows builds
- removes
Built locally from macOS (Apple Silicon) using cargo-zigbuild.