diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS
index d60f1c3dbe..94164648c2 100644
--- a/.github/CODEOWNERS
+++ b/.github/CODEOWNERS
@@ -1,2 +1,8 @@
 # Default owner for everything in the repo.
 * @Hmbown
+
+# AI code review is advisory and not wired through CODEOWNERS: GitHub CODEOWNERS
+# only accepts users and teams, not bots. @Hmbown stays the human code owner.
+#   - Claude: .github/workflows/claude-review.yml (GitHub Actions).
+#   - Codex/ChatGPT: the ChatGPT Codex cloud integration (chatgpt.com/codex ->
+#     connect GitHub -> enable Code review), authed by the ChatGPT subscription.
diff --git a/.github/workflows/claude-review.yml b/.github/workflows/claude-review.yml
new file mode 100644
index 0000000000..4f6dabe216
--- /dev/null
+++ b/.github/workflows/claude-review.yml
@@ -0,0 +1,81 @@
+name: Claude PR Review
+
+# Advisory AI code review by Claude (anthropics/claude-code-action) on every
+# non-draft PR. CODEOWNERS (@Hmbown) stays the human owner — this review posts
+# alongside it, it does not replace approval.
+#
+# Setup: add a CLAUDE_CODE_OAUTH_TOKEN repository secret
+#   1. Run `claude setup-token` locally (Pro/Max subscription) to mint a token.
+#   2. Settings -> Secrets and variables -> Actions -> New repository secret.
+# Until the secret exists the job no-ops with a notice (stays green), so this
+# workflow is safe to merge before the token is configured.
+
+on:
+  pull_request:
+    types: [opened, synchronize, reopened, ready_for_review]
+    branches: [master, main]
+
+concurrency:
+  group: claude-review-${{ github.event.pull_request.number }}
+  cancel-in-progress: true
+
+jobs:
+  claude-review:
+    name: Claude review
+    if: github.event.pull_request.draft == false
+    runs-on: ubuntu-latest
+    env:
+      HAS_OAUTH: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN != '' }}
+    permissions:
+      contents: read
+      pull-requests: write
+      id-token: write
+    steps:
+      - name: Skip when token is unset
+        if: env.HAS_OAUTH != 'true'
+        run: echo "::notice::CLAUDE_CODE_OAUTH_TOKEN is not set — skipping Claude review. Add the secret to enable it."
+
+      - name: Checkout repository
+        if: env.HAS_OAUTH == 'true'
+        uses: actions/checkout@v7
+        with:
+          fetch-depth: 1
+
+      - name: Claude code review
+        if: env.HAS_OAUTH == 'true'
+        uses: anthropics/claude-code-action@v1
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+          track_progress: true
+          prompt: |
+            REPO: ${{ github.repository }}
+            PR NUMBER: ${{ github.event.pull_request.number }}
+
+            You are reviewing a pull request against CodeWhale, a Rust workspace
+            (an agentic coding TUI/runtime). The PR branch is already checked out
+            in the current working directory.
+
+            Review the diff and report findings in this priority order:
+            1. Correctness bugs: logic errors, panics, unwrap/expect on fallible
+               paths, race conditions, incorrect error handling, off-by-one, and
+               non-exhaustive matches that could break compilation.
+            2. Provider/model/route safety (v0.8.65 EPIC #2608 invariant): a
+               provider-prefixed model string (e.g. `deepseek-ai/`, `deepseek/`,
+               `anthropic/`, `openai/`, `qwen/`) is a wire id or namespace hint,
+               never proof of provider selection. Flag any code that infers a
+               provider/model switch from such a prefix or from freeform prompt
+               text rather than from explicit user choice, config, Fleet policy,
+               capability requirements, or fallback policy.
+            3. Reuse and simplification: duplicated logic, dead code, needless
+               allocation/cloning, or reimplementing something the workspace
+               already provides.
+            4. Tests: missing coverage for new behavior and edge cases.
+            5. Security: secret handling, shell/exec policy, input validation.
+
+            Be specific and concise. Use inline comments for line-specific issues
+            and one top-level comment for the summary. Note genuinely good choices
+            briefly. Do not nitpick style that `cargo fmt` / `clippy` already
+            enforce.
+
+          claude_args: |
+            --allowedTools "mcp__github_inline_comment__create_inline_comment,Bash(gh pr comment:*),Bash(gh pr diff:*),Bash(gh pr view:*),Bash(git log:*),Bash(git diff:*)"