Verify a decision

{"actor":"judge","decided_at":"2026-08-25T12:55:30.316Z","decision":"review_revise","investigation_id":"10622728-729b-44eb-90d4-6ea9f25dae1a","new_score":0,"page_slug":"node-gyp-npm-supply-chain-compromise-june-2026","prev_score":0,"reason":"The page's core scaffolding -- campaign scale, the Phantom Gyp/binding.gyp mechanism, the Wave 1/Wave 2 timeline, and exact package and version figures -- is confirmed with high precision against multiple independently fetched vendor sources (claim_findings[0], [3], [6], [18]), and the raw disputed-plus-unverifiable rate is a low 12.5% (4 of 32 claims). However, two issues push this above a clean approval. First, nine of 32 findings are only 'partially_supported,' and several of them are not benign imprecision: claim_findings[7] shows the page's four-stage payload description conflicts with a materially different architecture in one of its own cited sources (ReversingLabs) without flagging the disagreement, and claim_findings[15] and [16] present specific persistence-mechanism names, timers, and AI-assistant file paths that could not be traced to the sources cited for them. Claim_findings[29] is an outright disputed timeline entry: the page's own cited source was published three days before the date the timeline attributes its reporting to. Second, and more consequential, the review identifies a high-priority coverage gap: while the body text correctly states that the genuine node-gyp package and its maintainers were not backdoored (claim_findings[5]), the page's title, slug, and entity name ('node-gyp npm Supply Chain Compromise') can be read as implying the legitimate, widely-depended-upon node-gyp tool itself was compromised. That framing risk is invisible to the disputed_pct metric, which only scores body-text claims, not title/entity naming. A second high-priority gap notes a possible unreported Microsoft/GitHub impact. Together these findings warrant revision -- clarifying the entity framing and resolving the timeline date and payload-description discrepancies -- rather than a clean approval or a denial, given that the substantive factual record otherwise holds up well.","score_delta":-14,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}