What problem does it solve? GitHub PR review workflows break in two subtle ways: Python thread-fetch scripts mis-decode GitHub API JSON under Windows locale encodings, and scanner findings (Codacy, Semgrep, CodeQL) get suppressed without exact rule IDs, leaving findings unresolved or incorrectly closed. This Skill adds evidence gates that catch both failure modes before they corrupt a review cycle. ## Core Features & Use Cases - Windows UTF-8 Gate: Forces UTF-8 mode (e.g., PYTHONUTF8=1) and strict decoding for Python-based thread fetchers so non-ASCII review text never silently truncates the thread inventory. - Scanner Suppression Gate: Requires mapping every check-run annotation to an exact scanner rule ID, tool, path, and line before prescribing any suppression, and demands a fresh remote analysis to prove the finding closed. - Pre-Delivery Verification: Re-checks each accepted fix against the original behavioral claim, effective tool configuration, evidence epoch, and exact reviewed head, with idempotency checks so only one review is posted. - Use Case: While addressing PR review comments on a Windows machine, a fetch script decodes GitHub JSON with the local code page and drops comments containing non-ASCII text; this Skill forces UTF-8 decoding and flags the incomplete inventory before you reply to a partial thread. ## Quick Start Ask the assistant to use gh-address-comments-extras when fetching PR review threads on Windows or before adding a suppression for a Codacy, Semgrep, or CodeQL finding.
