kxco-post-quantum 1.6.1 → 1.6.3

package/CHANGELOG.md

@@ -1,5 +1,54 @@
 # Changelog
 
+## 1.6.3
+
+Release plumbing only. No source change.
+
+1.6.2 shipped without its evidence bundle, because the upload step was gated on
+a `release` event that cannot occur: publish.yml creates the release with the
+default GITHUB_TOKEN, and GitHub raises no workflow events for actions taken
+with that token. The bundle attached by hand afterwards came from a
+push-triggered build, so it carried 855 subsampled vectors while the README
+states that 1,479 are in the downloadable bundle. Correctly labelled in the
+manifest, but not what the README describes.
+
+This release exists so a tag produces the bundle the README promises: the
+workflow now runs on `v*` tags, treats a tag as a full build rather than a
+subsampled one, and uploads without a human in the loop.
+
+## 1.6.2
+
+No change to the cryptography, the exports or the wire format. This release
+exists because the evidence describing them was wrong.
+
+**The interoperability report contradicted itself.** Every check wrote its
+failure sentence unconditionally while `ok` was computed separately, so a
+passing tamper control published `{"ok": true, "detail": "peer accepted a
+tampered signature"}`, and a passing byte-equality check published
+"deterministic signatures differ" followed by two identical values. 150 passing
+checks in the last bundle carried text that reads as a security failure. Every
+check now takes its sentence from its outcome.
+
+**The evidence bundle could not be built in CI, and never had been.** The
+workflow ran `npm ci` and nothing else: no ACVP vector fetch, no Java, no
+pinned Python peers, no liboqs image. Eight runs, both runtimes, all failed at
+the same step, so the only bundle an assessor could reach was hand-built from
+an uncommitted working tree. The workflow now does the setup its sibling always
+did, and the build starts from an empty directory rather than inheriting files
+from the run before it.
+
+**OpenSSL changed its default ML-DSA private key encoding in a patch release.**
+3.5.6 writes the 54-byte seed form, 3.5.7 the 4098-byte expanded form. FIPS 204
+allows either and this package still writes the seed form, which 3.5.7 continues
+to accept. The conformance test compared our bytes against whatever OpenSSL
+preferred, so it began failing on a change that was not ours. It now holds the
+byte-identity claim where OpenSSL writes seeds, and asserts that OpenSSL parses
+our seed form to the same seed and public key where it does not.
+
+**The dependency audit failed on its own regex**, matching an import specifier
+inside a string in an error message and reporting an unresolvable module named
+`priv`. Same 39 files from 12 entry points, same verdicts, no false failure.
+
 ## 1.6.0
 
 Additive. No existing export changes shape in a way a caller can observe, no

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "kxco-post-quantum",
-  "version": "1.6.1",
+  "version": "1.6.3",
   "description": "ML-DSA-65, ML-KEM-768 and SLH-DSA-SHA2-192s with key fingerprinting. OpenSSL 3.5 primitives on Node 24+, JavaScript elsewhere. 2,103 NIST ACVP vectors and 225 cross-implementation interop checks, 0 failed. Reproducible builds and SLSA provenance.",
   "keywords": [
     "post-quantum",