diff --git a/docs/caller-workflow.yml b/docs/caller-workflow.yml
index 178db6a..21fed07 100644
--- a/docs/caller-workflow.yml
+++ b/docs/caller-workflow.yml
@@ -1,8 +1,13 @@
 # Copy this to .github/workflows/code-review.yml in the target repository.
-# Change the two things marked CHANGE ME and nothing else.
+# For an SDK repository, change nothing. For anything else, read the two
+# CHANGE ME notes.
 #
 # This is the shape that works. It is not the obvious shape, and the comments
 # say why, because every deviation below was paid for once already.
+#
+# It should need NO edit at all for an SDK repository: every one of them names
+# its pull-request CI `Tests`, and every one of them is public and on the
+# MarketDataApp account, so `runs-on` is already right.
 
 name: Code review
 
@@ -17,11 +22,19 @@ name: Code review
 # notice. Do not add it back without reading `require_ci_green` first.
 on:
   workflow_run:
-    # CHANGE ME: the `name:` of every CI workflow whose result the review should
-    # wait for -- not the filename. Name them ALL: the gate looks at every check
-    # on the commit, so whichever finishes last is the one whose review actually
-    # proceeds. The earlier triggers skip cheaply, and a repeat on the same head
-    # skips too, so extra triggers are self-deduplicating rather than wasteful.
+    # LEAVE THIS ALONE. Every SDK repository names its pull-request CI `Tests`,
+    # which is what makes this file identical everywhere -- see
+    # docs/setup.md section 4a. It is the workflow's `name:`, never its filename.
+    #
+    # Both are named because the gate looks at EVERY check on the commit, so
+    # whichever finishes last is the one whose review actually proceeds; the
+    # earlier trigger skips cheaply, and a repeat on the same head skips too, so
+    # extra triggers are self-deduplicating rather than wasteful.
+    #
+    # Naming a workflow this repository does not have is harmless -- only sdk-py
+    # has a separate `Lint`; the others lint inside `Tests` -- because a name
+    # that never occurs never matches. If you add a CI workflow whose result
+    # should gate a review, add its name here.
     workflows: ["Tests", "Lint"]
     types: [completed]
   issue_comment:
diff --git a/docs/setup.md b/docs/setup.md
index a03b225..b3657f5 100644
--- a/docs/setup.md
+++ b/docs/setup.md
@@ -170,6 +170,33 @@ If you pin `uses:` to a branch or tag, pin `bot-ref` to the **same** ref.
 default is `main`. A mismatch fails with `invalid choice` on whichever command
 the older side lacks.
 
+### 3a. The CI workflow must be named `Tests`
+
+The trigger matches a workflow by its `name:`, so the name is an interface. All
+five SDK repositories now use `Tests`, which is what lets one caller file be
+copied without edits.
+
+They did not start that way — `Tests` in sdk-go, sdk-php and sdk-py, `CI` in
+sdk-js, `Pull Request` in sdk-java — and naming the wrong one **fails silently**:
+no review ever runs and nothing says why. sdk-java is the trap, because its
+`Main` workflow looks like the obvious candidate and only runs on pushes to the
+default branch.
+
+**Find the right workflow empirically, not from filenames.** Look at which
+checks actually appear on an open pull request:
+
+```bash
+sha=$(gh api repos/<owner>/<repo>/pulls/<n> --jq .head.sha)
+gh api "repos/<owner>/<repo>/commits/$sha/check-runs" --jq '.check_runs[].name'
+```
+
+then find which workflow produces them. Rename that workflow to `Tests`.
+
+**Renaming is safe.** Every required status check in these repositories is a
+**job** name, not a workflow name — `test (3.10)`, `Verify (JDK 17)`,
+`check (20.x)` — so branch protection is untouched. The only thing that moves is
+the label in the Actions list.
+
 ### 4. Add a policy, if the defaults are not what you want
 
 `.github/code-review/policy.yml` on the **base** branch. Absent keys keep the
diff --git a/tests/test_workflow.py b/tests/test_workflow.py
index 2c2874d..0526dec 100644
--- a/tests/test_workflow.py
+++ b/tests/test_workflow.py
@@ -534,3 +534,22 @@ def test_the_caller_template_still_names_its_secrets():
         "CODE_REVIEW_APP_PRIVATE_KEY",
         "CLAUDE_CODE_OAUTH_TOKEN",
     }
+
+
+def test_the_caller_template_needs_no_edit_for_an_sdk_repository():
+    """The workflow name is an interface, so it is standardised, not per-repo.
+
+    They diverged -- Tests in sdk-go/sdk-php/sdk-py, CI in sdk-js, Pull Request
+    in sdk-java -- and naming the wrong one fails SILENTLY: no review runs and
+    nothing says why. All five now use `Tests`, which is what lets this file be
+    copied without edits.
+    """
+    assert CALLER[ON]["workflow_run"]["workflows"] == ["Tests", "Lint"]
+    text = (ROOT / "docs/caller-workflow.yml").read_text()
+    assert "LEAVE THIS ALONE" in text
+    # The TRIGGER block specifically: `runs-on` still carries a CHANGE ME, for a
+    # repository that is private or outside the organisation, and that is right.
+    triggers = text[text.index("\non:"):text.index("\njobs:")]
+    assert "CHANGE ME" not in triggers, (
+        "the trigger block must need no edit for an SDK repository"
+    )