The modernization retains the existing public error API, synchronous construction, and one compiled ESM package. Unit regressions cover generated and supplied timestamps, capability changes, restricted clocks, JSON reconstruction, typed causes, and aggregate error graph traversal. No runtime adapter is installed by the library.
npm run checkverifies types, unit tests, lint, and packed-package behavior.npm run test:browserverifies the packed package in Chromium document and dedicated-worker realms.nix build '.#checks.x86_64-linux.default'runs routine hermetic verification throughverify.nix build '.#checks.x86_64-linux.runtime'selects the independentintegration-browserenvironment adapter.nix build '.#npm-package'builds the verbatim prepared archive.nix build '.#npm-package-unpacked'builds its exact unpacked view.
Checkout artifacts live under tmp/junit, tmp/coverage,
tmp/package-candidate, and tmp/integration-browser. Check derivations retain
a root report and an artifacts directory. Failed runs are not successful
evidence; support claims must identify the commands and environments actually
executed.
Executed on x86_64-linux with Node.js 24.15.0, TypeScript 5.9.3, MatrixAI lint 1.0.1, Playwright 1.59.1, and Nix-provided Chromium 147.0.7727.15:
| Surface | Result |
|---|---|
| Development activation | Standards submodule initialized on master, profile-selected skills materialized, direnv authorized, integrated shell activated. |
| Autofix | npm run lintfix passed across JS/TS, Markdown, shell, and Nix. |
| TypeScript | npm run typecheck passed with aligned NodeNext configuration. |
| Unit regressions | npm test -- --runInBand --coverage passed 28 tests in two suites. V8 coverage of the emitted clock module was 100% statements and branches; overall emitted-module statement coverage was 97.16%. |
| Packed package | Canonical manifest bytes, root/submodule exports, declarations without Node ambient types, and browser bundling without aliases passed. |
| Browser realms | Packed-package errors, timestamps, JSON reconstruction, inheritance, and cause-graph traversal passed in a Chromium document and dedicated worker. |
| Routine hermetic gate | nix build '.#checks.x86_64-linux.default' passed with retained verification reports and artifacts. |
| Runtime hermetic gate | nix build '.#checks.x86_64-linux.runtime' passed with retained browser cases and bundle-input metadata. |
| Producer outputs | Both prepared outputs built. Every archive file matched its unpacked counterpart, and the packed manifest matched the source byte-for-byte. Schema-v1 descriptors matched canonical runtime declarations and Node engines and reported dirty-checkout provenance without claiming a release revision. |
The prepared lint producer remained selected by Git tag v1.0.1 and resolved
commit bae774a6617b20d688de550d36c4e481fb8f9f69. Hermetic derivations used its
prepared archive through importNpmLock and the matching installation hook.
Installer behavior is not part of the errors test suite.
These are local Linux and Chromium results. No deployed Cloudflare, alternate browser engine, Windows, or macOS evidence is implied. At that modernization stage, no package was published or tagged. A later downstream integration run is separate from these errors-package proof nodes.
After the errors-only gates passed, an isolated copy of the container checkout was tested against the prepared unpublished errors archive. The copied dependency tree received the archive; only the copy had its browser clock alias and shim fixture removed. The actual container checkout, Git index, manifests, lockfile, and existing shim remained unchanged.
The isolated copy passed typechecking, 95 unit tests, packed-package checks, and the container-source import boundary check. Full browser compatibility was not established: all three browser cases stopped at bundling because the separate timer dependency, version 2.1.1, still imports the Node performance module. Removing the errors clock import does not make that transitive timer portable.
This diagnostic is not an installation test, a hermetic container proof, an updated consumer pin, or evidence of a released errors version. The errors manifest version remained 2.1.3 during that uncommitted modernization. That diagnostic did not authorize a consumer update or release.
The Node 24 minimum is released as a major version. The portable clock and verbatim Nix archive contract are retained. The final 3.0.0 routine and browser gates passed before release preparation. The archive and unpacked outputs also built successfully. It is not published to npm. The legacy tag-triggered npm publishing workflow is removed to keep Git-only releases from invoking registry publication.
The shared consumer now resolves and temporarily normalizes nested Git dependency edges for npm's clean-install validation, then restores canonical constraints. The matching Nix importer normalizes those same temporary edges. Producer archives and published dependency metadata remain verbatim. The redundant prepack hook is removed; prepare and explicit Nix builds retain the existing compilation boundary. Lint uses an explicit SSH Git URI at the unchanged 1.0.1 tag and revision. Existing 3.0.0 tags remain immutable.
This patch carries the common npm-script development workflow, explicit lint backend provisioning, identical generic lint configuration, and removal of the obsolete GitHub workflows. All six libraries passed their lintfix commands in the aligned Nix development environments before this dependency-pin update. The final release graph receives a prepared archive build; broad runtime suites are not repeated. Existing release tags remain immutable. No npm publication is performed.
Selected released family inputs:
- lint 1.0.2 at commit 074b1078b2da0634e69f33852b32b32ea60d2e61.