diff --git a/.github/workflows/align-dist-tags.yml b/.github/workflows/align-dist-tags.yml
new file mode 100644
index 0000000000..c9e9511088
--- /dev/null
+++ b/.github/workflows/align-dist-tags.yml
@@ -0,0 +1,105 @@
+name: Align dist-tags
+
+# Manual remedy for the `@next == @latest` alias invariant (CLAUDE.md).
+#
+# publish.yml aligns @next automatically after every publish (DIST-TAG-01, #391). This
+# workflow exists for the cases that path cannot cover: a release published BEFORE that
+# fix landed, or an alignment that failed transiently. Doing it by hand needs
+# `npm dist-tag add`, which on a 2FA-protected account prompts EOTP for a one-time
+# password — so it cannot be scripted, and in practice it does not get done.
+#
+# The NPM_TOKEN secret is 2FA-exempt (that is what automation tokens are for), so CI can
+# do what an interactive human cannot do unattended.
+#
+# Scope is deliberately narrow: this moves a dist-tag and nothing else. It cannot publish.
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version to point @next at (default: whatever @latest currently is)'
+        required: false
+        type: string
+
+permissions:
+  contents: read
+
+jobs:
+  align:
+    name: Point @next at @latest
+    runs-on: ubuntu-latest
+    environment: npm-publish
+    timeout-minutes: 5
+    steps:
+      - name: Resolve target version
+        id: target
+        # `inputs.version` is attacker-controlled on a workflow_dispatch. GitHub expands
+        # ${{ }} BEFORE bash parses the script, so interpolating it inline would let a
+        # crafted input close the quote and run arbitrary commands. Passing it through
+        # `env:` keeps it a shell VALUE that is never parsed as script.
+        env:
+          INPUT_VERSION: ${{ inputs.version }}
+        run: |
+          LATEST=$(npm view @nforma.ai/nforma dist-tags.latest 2>/dev/null || echo '')
+          TARGET="${INPUT_VERSION}"
+          [ -z "$TARGET" ] && TARGET="$LATEST"
+          if [ -z "$TARGET" ]; then
+            echo "::error::could not resolve a target version (npm view returned nothing and no input was given)"
+            exit 1
+          fi
+          # Whitelist the shape as well as escaping it: only a bare stable semver may
+          # reach `npm dist-tag`. This rejects anything with shell metacharacters
+          # outright rather than relying on quoting alone, and doubles as the
+          # prerelease refusal — under the alias policy @next == @latest, a prerelease
+          # holding @next would silently install for every user.
+          # Whole-string check, not line-wise: `grep` is line-oriented, so a value like
+          # $'0.0.1\ntouch x' passes an anchored pattern on its FIRST line while still
+          # carrying a newline — which then injects extra keys into $GITHUB_OUTPUT.
+          # Stripping the legal alphabet and requiring nothing remains catches that.
+          if [ -n "${TARGET//[0-9.]/}" ] || ! printf '%s' "$TARGET" | grep -Eq '^[0-9]+\.[0-9]+\.[0-9]+$'; then
+            echo "::error::refusing target '${TARGET}' — expected a bare stable semver (x.y.z); prereleases and any other shape are rejected"
+            exit 1
+          fi
+          echo "target=${TARGET}" >> "$GITHUB_OUTPUT"
+          echo "latest=${LATEST}" >> "$GITHUB_OUTPUT"
+          echo "Target: ${TARGET} (current @latest: ${LATEST:-unknown})"
+
+      - name: Align @next
+        env:
+          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
+          TARGET: ${{ steps.target.outputs.target }}
+        run: |
+          if [ -z "${NPM_TOKEN}" ]; then
+            echo "::error::NPM_TOKEN secret is not set — this workflow cannot authenticate"
+            exit 1
+          fi
+          # npm does NOT read a bare NPM_TOKEN env var; auth comes from .npmrc. That exact
+          # mistake is what made every release's alignment silently 401 (DIST-TAG-01).
+          printf '//registry.npmjs.org/:_authToken=${NPM_TOKEN}\n' > .npmrc
+          trap 'rm -f .npmrc' EXIT
+          npm dist-tag add "@nforma.ai/nforma@${TARGET}" next
+
+      - name: Verify the invariant holds
+        env:
+          TARGET: ${{ steps.target.outputs.target }}
+        run: |
+          TAGS=$(npm view @nforma.ai/nforma dist-tags --json 2>/dev/null || echo '{}')
+          echo "$TAGS"
+          LATEST=$(printf '%s' "$TAGS" | node -p "JSON.parse(require('fs').readFileSync(0,'utf8')).latest || ''")
+          NEXT=$(printf '%s' "$TAGS" | node -p "JSON.parse(require('fs').readFileSync(0,'utf8')).next || ''")
+          # Unlike publish.yml's verify — which must never fail a job whose package is
+          # already published — a manual alignment that did not align is a failure. A
+          # green run that changed nothing is the outcome worth preventing here.
+          if [ -z "$LATEST" ] || [ -z "$NEXT" ]; then
+            echo "::error::could not read dist-tags back — alignment unverified"
+            exit 1
+          fi
+          if [ "$NEXT" != "$TARGET" ]; then
+            echo "::error::@next is ${NEXT}, expected ${TARGET} — alignment did not take"
+            exit 1
+          fi
+          if [ "$LATEST" != "$NEXT" ]; then
+            echo "::error::invariant still broken — latest=${LATEST} next=${NEXT}"
+            exit 1
+          fi
+          echo "invariant OK — next == latest == ${LATEST}"
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 9d6772d130..f8ebb6dc33 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -27,6 +27,7 @@ Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 ## [Unreleased]
 
 ### Added
+- feat(release): **`Align dist-tags` — a manually dispatchable workflow that repairs `@next` drift without a one-time password.** `publish.yml` aligns `@next` automatically after every publish since #391, but that cannot help a release published *before* that fix, or an alignment that failed transiently. Doing it by hand needs `npm dist-tag add`, which on a 2FA-protected account fails with `EOTP` and demands an authenticator code — so it cannot be scripted, and in practice it simply does not get done (0.44.3 shipped with `next` still on 0.44.2). The `NPM_TOKEN` secret is 2FA-exempt, which is what automation tokens are for, so CI can do unattended what an interactive human cannot. Scope is deliberately narrow: it moves a dist-tag and cannot publish. It refuses a prerelease target (a prerelease holding `@next` would silently install for everyone under the alias policy), and — unlike `publish.yml`'s verify step, which must never fail a job whose package is already out — this one **fails** if the tag did not move, because a green manual run that changed nothing is the exact outcome worth preventing. Each `run:` block was executed against a stubbed `npm view` across all four branches (aligned / not-moved / unreadable / prerelease) rather than reviewed by eye.
 - feat(quorum): **timeout records now say which config produced the window (I5).** `resolveTtfbTimeout` / `resolveInterChunkCeiling` each compute a `source` (`per-slot` / `default` / `idle-budget` / `legacy-stall`) and `recordTelemetry` was dropping both it and the resolved values, so a TIMEOUT in `quorum-rounds-*.jsonl` could be seen but never attributed — the number alone never answers "did *my* config do this?", since today's per-slot value can be tomorrow's default. Records now carry `ttfb_timeout_ms`, `ttfb_source`, `inter_chunk_ceiling_ms` and `inter_chunk_source`, always present (null when unknown) so the schema stays stable for analysis. Passed as a trailing options object rather than a 15th positional parameter. 4 tests drive the **real child** and read the actual log, so the fields are proven to survive the whole dispatch path; mutation-proven. Accepted as improvement I5 by the 2026-08-07 quorum.
 
 ### Fixed