diff --git a/harness/tests/README.md b/harness/tests/README.md
index 796688e..610f8f4 100644
--- a/harness/tests/README.md
+++ b/harness/tests/README.md
@@ -1,5 +1,45 @@
 # Canonical harness tests
 
+## Installation-root prerequisite
+
+Before executing the suite, provide `HARNESS_ROOT` in the process environment.
+It names the installation root containing `harness/` and `repositories/`, not
+the `harness/` directory itself. For the current installation, configure the
+current agent's `$HERMES_HOME/.env` with:
+
+```dotenv
+HARNESS_ROOT=/opt/data/gitea-agent
+```
+
+The tests consume the environment supplied by Hermes; they do not open or
+source `.env` files, search other profiles, or implement a second configuration
+loader. When running outside Hermes, supply the variable explicitly:
+
+```sh
+export HARNESS_ROOT=/opt/data/gitea-agent
+python3 -B "$HARNESS_ROOT/harness/tests/run.py"
+```
+
+On another system, replace the value with that installation's absolute root.
+The managed-lock subprocess contention test requires a nonempty, existing
+absolute directory and compares its resolved location with the installation
+containing the test file. Missing, empty, relative, nonexistent, non-directory,
+or different-installation values raise an explicit `HARNESS_ROOT` error before
+creating a lock or starting the contention subprocess. There is no fallback to
+`Path.cwd()` or to the test-file-derived root. Only the validated import root
+is passed as a subprocess argument; no `.env` contents are read or serialized.
+This test-only prerequisite does not change the production lock implementation
+or the canonical runner's source-bound collection behavior; `--collect-only`
+does not require the variable.
+
+An external-cwd regression runs only the contention test from a temporary
+working directory, with `PYTHONPATH` removed and an explicit fixture
+`HARNESS_ROOT`, using isolated Python and bounded subprocess timeouts. Invalid
+configuration coverage checks that neither lock acquisition nor subprocess
+execution occurs. It does not recursively invoke the entire suite.
+
+## Running the canonical suite
+
 The canonical runner is anchored to its own source tree and works from any
 current directory. From the repository root:
 
diff --git a/harness/tests/common/test_managed_lock.py b/harness/tests/common/test_managed_lock.py
index e026dc0..73abf69 100644
--- a/harness/tests/common/test_managed_lock.py
+++ b/harness/tests/common/test_managed_lock.py
@@ -241,7 +241,61 @@ class ManagedLockTests(unittest.TestCase):
             after = path.stat()
             self.assertEqual((before.st_ino, before.st_size, before.st_mode), (after.st_ino, after.st_size, after.st_mode))
 
+    def test_lock_contention_from_external_cwd_without_pythonpath(self):
+        root = Path(__file__).resolve().parents[3]
+        script = """
+import sys
+import unittest
+sys.path.insert(0, sys.argv[1])
+suite = unittest.defaultTestLoader.loadTestsFromName(
+    'harness.tests.common.test_managed_lock.ManagedLockTests.'
+    'test_other_process_cannot_acquire_while_context_alive')
+result = unittest.TextTestRunner().run(suite)
+raise SystemExit(not result.wasSuccessful())
+"""
+        env = os.environ.copy()
+        env.pop("PYTHONPATH", None)
+        env["HARNESS_ROOT"] = str(root)
+        with tempfile.TemporaryDirectory() as cwd:
+            result = subprocess.run(
+                [sys.executable, "-I", "-B", "-c", script, str(root)],
+                cwd=cwd, env=env, text=True, capture_output=True, timeout=30)
+        self.assertEqual(result.returncode, 0, result.stdout + result.stderr)
+
+    def test_invalid_harness_root_stops_before_lock_and_subprocess(self):
+        with tempfile.TemporaryDirectory() as other:
+            file_path = Path(other) / "not-a-directory"
+            file_path.write_text("fixture")
+            for value in (None, "", "relative", other,
+                          str(Path(other) / "missing"), str(file_path)):
+                with self.subTest(value=value), patch.dict(os.environ), \
+                        patch(__name__ + ".ManagedLock") as lock, \
+                        patch(__name__ + ".subprocess.run") as run:
+                    if value is None:
+                        os.environ.pop("HARNESS_ROOT", None)
+                    else:
+                        os.environ["HARNESS_ROOT"] = value
+                    run.return_value.stdout = "blocked"
+                    error = None
+                    try:
+                        self.test_other_process_cannot_acquire_while_context_alive()
+                    except Exception as exc:
+                        error = exc
+                    self.assertIsInstance(error, RuntimeError)
+                    self.assertIn("HARNESS_ROOT", str(error))
+                    lock.assert_not_called()
+                    run.assert_not_called()
+
     def test_other_process_cannot_acquire_while_context_alive(self):
+        value = os.environ.get("HARNESS_ROOT")
+        if not value:
+            raise RuntimeError("HARNESS_ROOT must be configured")
+        root = Path(value)
+        if not root.is_absolute() or not root.is_dir():
+            raise RuntimeError("HARNESS_ROOT must be an existing absolute directory")
+        root = root.resolve()
+        if root != Path(__file__).resolve().parents[3]:
+            raise RuntimeError("HARNESS_ROOT does not match this test installation")
         with tempfile.TemporaryDirectory() as d:
             path = Path(d) / "process.lock"
             script = """
@@ -255,7 +309,7 @@ except LockTimeoutError:
     print('blocked')
 """
             with ManagedLock(path):
-                result = subprocess.run([sys.executable, "-c", script, str(path), str(Path.cwd())], text=True, capture_output=True, check=True)
+                result = subprocess.run([sys.executable, "-c", script, str(path), str(root)], text=True, capture_output=True, check=True, timeout=30)
             self.assertEqual(result.stdout.strip(), "blocked")