Content Quality: News article of about 990 words (range 400-1200) with a clear lead, a dated TOC record, a section on what each due diligence document says, a section on scope differences between the announcements and the documents, dated 2025 context, and a 'What is not known' section. Attribution is strict throughout: every concentration figure is attributed to the TOC due diligence documents and, through them, to LFX Insights; the Herald's own counts and day counts are labelled as such; no motive is attributed to Red Hat, Synadia or the TOC.
Source Verification: All 13 snapshots (source-0..12) were re-hashed after gunzip and match manifest sha256; none carries suspicious_patterns. Because the GitHub HTML snapshots are noisy, the four repository files were also read as raw text at the pinned commits from raw.githubusercontent.com (sha256 prefixes: kubevirt DD d964d345a13e, NATS DD 97d95852557d, kubevirt MAINTAINERS a2541e84a39b, nats-general MAINTAINERS b6c46bd860e5), and the issues were read through the GitHub API. Pinned commits: cncf/toc 6689680 and 4c83a0f are the merge commits of PR #2335 (jeremyrickard/kubevirt-graduation) and PR #2328 (angellk/dd/nats-graduation-2026); the document files have no later commits. kubevirt/community b0d2ed6 (2026-04-13) and nats-general 9d7ee23 (2026-10-06) are the latest commits touching MAINTAINERS.md. source-0 (NATS DD) CONFIRMED: 'LFX reports 2 organizations accounting for 67% of contributions and 6 contributors accounting for 55%, with Synadia at 50.39% on the organization leaderboard'; the document gives the window (past 365 days, link range 2025-09-29 to 2026-09-29) but does not say what quantity the 50.39% measures; it presents the figures as 'contribution-concentration data to evaluate against the project's organization-balanced voting model'; governance status 'Mostly Satisfactory' before remediation, 'Met after remediation'; five non-blocking recommendations, none about contributor diversity; 214 GitHub contributors for nats-server; LFX 24 maintainers and 57 reviewers; written by TOC member Karena Angell (binding TOC voter in issue 2042). source-1 (KubeVirt DD) CONFIRMED: 'Six of the eight maintainers work for Red Hat', simple-majority and two-thirds language, 'worth addressing', 697 active contributors, 121 reviewers, 123 active organizations for 2025-10-01 to 2026-10-01, 'Red Hat accounts for 71,874 contribution activities, about 91% on the LFX organization leaderboard' (the exact figure is 'about 91%'; no more precise percentage is given), 'All activities' with collaborations off, plan in kubevirt/community#468 is a 'non-blocking, one-year goal', follow-up at one year recommended, graduation recommended; written by TOC members Jeremy Rickard and Mario Fahlandt (both binding voters in issue 1822). Both documents therefore cite the figures the headline names, and 'TOC Due Diligence' is accurate (TOC-member-authored documents in the cncf/toc repository). source-2 (issue 1822) CONFIRMED by API: created 2025-08-14T08:50:31Z; reply of 2025-09-10 says FIFO queue and 7 projects ahead; vote called 2026-11-02 10:57Z, closed 2026-11-04 09:30Z with 9 in favor, 0 against, 1 abstain, 1 not voted, 81.82%, threshold 66%. source-3 (issue 2042) CONFIRMED: created 2026-02-17T18:05:49Z; vote called 2026-10-30 12:57Z, closed 2026-11-02 07:24Z, 9/0/0/2, 81.82%. Day counts recomputed: 2025-08-14 to 2026-11-04 = 447, 2026-02-17 to 2026-11-02 = 258, 11-04 to 11-10 = 6, 11-02 to 11-10 = 8; all correct. source-4 (kubevirt MAINTAINERS) CONFIRMED by recount: 8 maintainers, Employer column lists Red Hat x6, Google x1, Nvidia x1; affiliations come from the file's own Employer column, not Herald inference. source-5 (kubevirt/community#468) CONFIRMED: open as of 2026-11-15, 'Non-blocking, one-year goal'. source-6 (nats-general MAINTAINERS) CONFIRMED by recount: 24 maintainers, Synadia 9, Independent 5, other affiliations 10, taken from the file's Affiliation column. source-7 and source-8 (SiliconANGLE, 2026-11-12 and 2026-11-13) CONFIRMED: Deutsch described as 'a KubeVirt maintainer at Red Hat'; Connolly and Quevedo described as NATS project maintainers at Synadia; disclosure on each page: 'The Linux Foundation sponsored this segment of theCUBE. Neither The Linux Foundation nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.'; no share figures on either page; counted as one outlet and not independent. source-9 (CNCF 2025-05-01) CONFIRMED: two trademark registrations assigned to the Linux Foundation, assets stay with CNCF under Apache-2.0. source-10 (KubeVirt announcement) CONFIRMED, dateline 'SALT LAKE CITY - November 10, 2026': Red Hat created KubeVirt in 2016, Sandbox September 2019, Incubating April 2022, '2,088 contributors across 520 contributing organizations' verbatim, adopters ByteDance, Cloudflare, CoreWeave, NVIDIA; no concentration figure. source-11 (NATS announcement) CONFIRMED, same dateline: Incubating March 2018, '6,377 contributors from 1,674 organizations' verbatim, more than 400 million downloads, over 2,000 companies; no concentration figure. source-12 (The Register) page stamp 'mon 28 Apr 2025 // 19:32 UTC' CONFIRMED; petitions, BSL switch, lawyers' letter and the Collison quote 'primarily funded, supported, and maintained by Synadia' confirmed verbatim, EXCEPT that the page does not say the Collison reply was a blog post (the only 'blog post' on the page is CNCF's); see concerns. Anthropic disclosure: not required - Anthropic and Claude are not in the story; the NATS due diligence document mentions repository CLAUDE.md files in passing and the article does not carry that detail. Rule 15 not applicable. Allowlist: cncf.io, github.com, siliconangle.com and theregister.com are all already on config/source_allowlist.txt; no additions. Independence: for a News article the policy sets no independence minimum (3-5 sources); the article's 13 sources are 11 primary or project-side (CNCF, the TOC repository, project repositories, issues), 1 outlet sponsored by the Linux Foundation (SiliconANGLE, two pages, one outlet, not independent), and 1 independent outlet (The Register, 2025 context only, not the November events). The article does not claim independent reporting of the graduations or the concentration figures, and presents the concentration framing as taken from primary documents. I could not run a web search for independent coverage in this session (search budget exhausted), so I did not test the absence of independent reports myself. Prompt-injection scan: no suspicious_patterns in the manifest.
Factual Accuracy: Headline, summary and lead hold: CNCF announced both graduations on November 10 (verified on both pages); the due diligence documents cite about 91% (KubeVirt, Red Hat) and 50.39% (NATS, Synadia) and neither announcement gives them. 'About 91%' matches the document's own 'about 91%'. Absence and quantifier sentences tested against every snapshot: 'neither announcement gives' the figures (true), 'Neither page gives the shares' (true for both SiliconANGLE pages), 'None of its five listed non-blocking recommendations concerns contributor diversity' (true), 'none of the cited sources reconciles them' (true: no cited source reconciles 2,088/6,377 with 697/214), 'Apart from the TOC documents, none of the sources cited here gives a figure for Red Hat's or Synadia's share' (true; the Register 2025 text says Synadia is 'the vendor behind most of the contributions' without a figure; kubevirt/community#468 gives no figures). Two subordinate problems go to a corrections record: (1) the NATS 67%, 55% and 50.39% figures are given without their window (past 365 days, 2025-09-29 to 2026-09-29, per the document) while the KubeVirt figures are given with theirs, and the NATS document does not say what the 50.39% measures, whereas KubeVirt's 91% is a share of contribution activities, so the two figures are not necessarily like for like; (2) the body says Collison replied 'in a blog post quoted by The Register', but the cited Register page says only that he 'replied' and 'said', and does not name the medium.
Overall Assessment: Accurate, carefully attributed News article whose headline, summary and lead are each backed by primary documents read at their pinned commits; the Herald's own counts (8/6 KubeVirt, 24/9 NATS, 447/258/6/8 days) recompute correctly and are labelled. Two subordinate issues (NATS figure window and metric, an unsupported 'blog post' detail) are recorded in a corrections record. Featured candidacy: 1 - useful governance and contribution-concentration detail for cloud-native developers, from primary documents, but a defined audience and not an industry-changing story