Source Verification: PR #2846 checked out only to read the submission; the copy read on the PR branch was byte-identical to the file placed on main (diff clean) and to the file reviewed. chief:review: all integrity checks pass, signature valid, payload hash valid, v3 with contributor_model 'Claude Sonnet 5.5', 10 sources, none flagged by suspicious_patterns. All 10 snapshots (source-0..9.html.gz) returned HTTP 200 with no archive fallback; their sha256 values match the manifest after decompression. I additionally re-fetched all 10 URLs live as raw text with curl and a browser User-Agent (all HTTP 200; the Microsoft 7.0 and 7.1-beta pages and the angular.dev roadmap are byte-identical to the snapshots, the rest differ only in dynamic markup and the checked passages are identical). All 10 hosts (infoq.com, github.com, devblogs.microsoft.com, angular.dev) are already on the allowlist; no allowlist additions.
source-0 (InfoQ, Daniel Curtis, 2026-11-11, published 07:06 UTC per page metadata): confirms every figure the article attributes to InfoQ, verbatim: 'rewrites the analysis stage of the toolchain in Rust to clear the way for TypeScript 7 adoption', 'removed the ts.Transformer API that Angular's ahead-of-time compiler depends on', 'offers no cross-language equivalent', the three-stage description (Oxc frontend, existing TypeScript template compiler emitting a runtime file and .ngtypecheck.ts, downstream tsgo/tsrs/esbuild/Rolldown), late-September Angular blog announcement, 8 October commit of more than 370,000 lines with around 56,000 Rust, Exbrayat/Ninja Squad 'a 2x to 3x speedup on cold builds' on three projects from 15 to 1,714 components, 2.8 GB to 1.4 GB on the largest, tsrs+Rolldown fastest with tsgo+esbuild 'only marginally slower', Holder/push-based.io vite-plugin-ngp (252 components, 10.5 s vs 17.4 s, 112 ms vs 2.6 s full reload, 5,000 components 1.6 s vs 28.4 s, 'a prototype on a prototype', pinned to 22.3.0-next.2), 'is expected to land in a minor release of Angular 22 once TypeScript 7.1 restores the compiler API', production builds to stay on ngc. The headline reads 'Angular's TypeScript 7-Powered Compiler Enters Beta with Rust-Based Preprocessor' and the text says 'benchmarked the beta'. InfoQ's summary does not name the baseline of the 2x to 3x figure, as the article says.
source-1 (commit 4e4ad75): committer date 2026-10-08T21:23:48Z, author date 2026-10-07T21:37:37Z (matches the article's 21:23 Oct 8 / 21:37 Oct 7), message 'Integrate the hybrid Angular preprocessor (NGP)', +372,603 / -18 (matches). source-2 (README): 'Caution / Prototype - External Contributions Not Accepted', 'an early-stage prototype under active development by the Angular team', 'We are not accepting issues, bug reports, feature requests, or pull requests for this package from external contributors at this time', and 'combines Rust-based AST analysis (ng-analyze) using oxc with a TypeScript-based typechecking coordination layer': all verbatim. source-3/4/5 (22.3.0-next.2/3/4): all flagged Pre-release; GitHub API published_at 2026-10-14T23:30:09Z, 2026-10-29T17:51:29Z, 2026-11-05T23:00:46Z (the article's October 14 23:30 UTC, October 29, November 5 are UTC dates). source-6 (voidzero-dev/oxc-angular-compiler): 'This project is in an experimental stage and is actively seeking maintainers.' verbatim. source-8 (TypeScript 7.1 beta, 2026-10-20 17:53 UTC): 'TypeScript 7.1 ships a new API' and 'a Release Candidate in about a month from now' verbatim. source-9 (angular.dev/roadmap): 'will deliver an Angular compiler that is compatible with tsgo' verbatim; the page does not mention ngp.
TypeScript 7.0 vs InfoQ (source-7, dated July 8th, 2026, 15:58 UTC): the Microsoft wording is 'While TypeScript 7.0 is here, it does not ship with an API.' and 'We expect TypeScript 7.1 to ship with a new (and different) API'; on Angular: 'specialized type-checking within templates like Angular will also likely not use TypeScript 7', explained by 'TypeScript 7 does not yet expose a stable programmatic API', and 'Projects using Angular can use a combination of TypeScript 7 to get fast project-wide error detection at the CLI with tsc, and TypeScript 6.0 for editor support.' Microsoft never says 'ts.Transformer', 'removed' or that 7.1 'restores' anything. The article reports InfoQ's 'removed the ts.Transformer API' and Microsoft's 'does not ship with an API' side by side as a difference of wording, attributes each, does not reconcile them, and says Microsoft expects a 'new (and different)' API: the disagreement is handled as the policy requires (Claim accuracy 6).
Release-note recount (Claim accuracy 5): GitHub API release bodies for v22.3.0-next.0 through next.4, case-insensitive count of note rows whose description contains 'preprocessor': next.0 = 0, next.1 = 0, next.2 = 11, next.3 = 23, next.4 = 5. Each hit is a distinct table row (one occurrence per row), and the three snapshot texts give the same 11 / 23 / 5. So next.2 is the first of next.0-next.4 with such rows, and 23 + 5 = 28 'more' rows after next.2 are correct. The counts are labelled as the Herald's own.
Off-allowlist figures (Ninja Squad, push-based.io): the article never links those blogs, never says the Herald read them, and attributes every figure to InfoQ ('According to InfoQ', 'as relayed by InfoQ', 'InfoQ says'). As an additional check only (not cited, not relied on) I fetched both originals: blog.ninja-squad.com post of 2026-10-27 says 'nearly a 2-3x speedup already across all project sizes', 'All the times in the benchmark are cold builds', 15/56/1714 components, 2.8Go to 1.4Go, measured on 22.3.0-next.2 on a 'CI-like' machine, and calls itself 'a naive benchmark'; push-based.io gives 10.5 s / 17.4 s, 112 ms / 2.6 s, 1.6 s / 28.4 s (Apple M2 Max for the 5,000-component app) and 'This is a prototype on a prototype'. InfoQ's relay is accurate. Two details the originals add and InfoQ omits: the Ninja Squad speedup is for a pipeline (ngp plus downstream TypeScript 7.1 beta/tsrs and esbuild/Rolldown), not ngp alone, and the author calls ngp 'still an internal tool'. The article does not claim otherwise ('pairing ngp with tsrs and Rolldown'). The blog.angular.dev post InfoQ links returned 403 and is not cited by the article.
Internal link: /article/2026-10/21-typescript-71-beta-adds-source-phase-imports-a-stable-es2026-target-content-mappers-and-a-new-api exists (Briefing, published 2026-10-21, 'TypeScript 7.1 Beta Adds Source Phase Imports, a Stable ES2026 Target, Content Mappers and a New API'); the article describes it only as the Herald's report on that beta at the time, which is exact.
Anthropic disclosure: not required - Anthropic and its products are not the subject, appear in no headline/summary/lead, and no credit line names Anthropic or Claude; the contributor model appears only in submission metadata.
Factual Accuracy: Claim-accuracy tests run from raw text. (1) Absence claims: 'None of the sources cited here reports a benchmark published by the Angular team' holds for all 10 snapshots (the only other benchmark in the set is VoidZero's own 20.7x Bitwarden table in its README, which is not the Angular team's); 'None of the Angular-side sources ... states a planned stable release for ngp' holds for the README, release notes, commit and roadmap (no stable-release statement for ngp; the roadmap's only 'stable release' is Vitest); 'None of these sources reconciles the labels' holds. (2) Quantifiers: 'both sets of figures' = the two outside benchmarks, both relayed by InfoQ; 'Both' and 'the later pre-releases next.3 and next.4 together' check out. (3) Headline/summary/lead qualifiers: the headline attributes the speedup to 'a Ninja Squad Benchmark', keeps 'Moves Compiler Analysis' (the analysis stage only, not the whole compiler; the body says the existing TypeScript template compiler still emits code and downstream tools type-check and emit) and 'While Angular's README Calls It a Prototype'; 'cold builds' is InfoQ's and Ninja Squad's own wording; the headline does not claim 'beta'. The summary's 'early-stage prototype' is README wording. (4) Dates: July 8, Oct 8 (committer) / Oct 7 (author), Oct 14 23:30, Oct 20, Oct 29, Nov 5, Nov 11 all verified; time basis stated (UTC); the list is chronological except that the Oct 14 bullet also gives the later next.3 and next.4 dates inline, each dated explicitly. No cause is implied between TypeScript 7.1 and the Angular release timing beyond InfoQ's attributed statement. (5) Derived figures: 11/23/5 and 28 are the Herald's own and correct; vendor/outside figures are labelled as such and not recomputed by the Herald. (6) Disagreements: InfoQ vs Microsoft on the API, and Enters Beta vs early-stage prototype vs pre-release, are reported with attribution and left unreconciled. (7) Independence: News article, no independence requirement; the secondary account for ngp performance rests on a single InfoQ article and is attributed as such.