diff --git a/e2e/wrap/run.py b/e2e/wrap/run.py
index 0120adc00..c28a8b1a9 100644
--- a/e2e/wrap/run.py
+++ b/e2e/wrap/run.py
@@ -790,6 +790,11 @@ def verify_vscode_wrap(base_env: dict[str, str], project_dir: Path) -> None:
             f'"http://127.0.0.1:{port}{project_prefix}"' in configured,
             "VS Code wrap should configure the project-scoped proxy URL",
         )
+        assert_true(
+            f'"github.copilot.advanced.debug.overrideCapiUrl": '
+            f'"http://127.0.0.1:{port}{project_prefix}"' in configured,
+            "VS Code wrap should route Copilot Chat generation through Headroom",
+        )
         assert_true(
             '"github.copilot.advanced.debug.overrideAuthType": "token"' in configured,
             "VS Code wrap should configure token auth",
@@ -849,8 +854,8 @@ def verify_vscode_claude_wrap(base_env: dict[str, str], project_dir: Path) -> No
             "VS Code Claude wrap should configure the project-scoped Anthropic URL",
         )
         assert_true(
-            configured["env"]["ENABLE_TOOL_SEARCH"] == "true",
-            "VS Code Claude wrap should retain Claude Code tool deferral",
+            configured["env"]["ENABLE_TOOL_SEARCH"] == "false",
+            "VS Code Claude wrap should disable tool deferral for webview compatibility",
         )
         assert_true(configured["env"]["KEEP"] == "yes", "Existing Claude env must remain")
         assert_true(str(settings_path) in output, "Wrap output should identify Claude settings")
diff --git a/headroom/cli/init.py b/headroom/cli/init.py
index 8a153ee22..2b6f4bf4c 100644
--- a/headroom/cli/init.py
+++ b/headroom/cli/init.py
@@ -41,6 +41,7 @@ from headroom.install.runtime import (
 from headroom.install.state import ManifestError, load_manifest, save_manifest
 from headroom.install.supervisors import start_supervisor
 from headroom.providers.claude import TOOL_SEARCH_DEFAULT, TOOL_SEARCH_ENV
+from headroom.providers.claude.runtime import TOOL_SEARCH_FOUNDRY_DEFAULT
 from headroom.providers.codex.install import codex_uses_chatgpt_auth
 from headroom.providers.codex.threads import retag_to_headroom
 
@@ -181,7 +182,12 @@ def _ensure_claude_hooks(path: Path, profile: str, port: int) -> None:
     # all into its context window — overflowing it (breaks sub-agent spawns,
     # forces constant compaction). Keep deferral on; respect a user-set value.
     # Shares the TOOL_SEARCH_* constants with `wrap` and `install`.
-    env_map.setdefault(TOOL_SEARCH_ENV, TOOL_SEARCH_DEFAULT)
+    tool_search_default = (
+        TOOL_SEARCH_FOUNDRY_DEFAULT
+        if os.environ.get("CLAUDE_CODE_USE_FOUNDRY")
+        else TOOL_SEARCH_DEFAULT
+    )
+    env_map.setdefault(TOOL_SEARCH_ENV, tool_search_default)
     payload["env"] = env_map
 
     hooks = dict(payload.get("hooks") or {}) if isinstance(payload.get("hooks"), dict) else {}
diff --git a/headroom/cli/wrap.py b/headroom/cli/wrap.py
index 8f1971a13..ee91a14a3 100644
--- a/headroom/cli/wrap.py
+++ b/headroom/cli/wrap.py
@@ -451,6 +451,8 @@ def _resolved_tool_search_mode(flag_value: str | None) -> str:
     existing = os.environ.get(_TOOL_SEARCH_ENV)
     if existing is not None:
         probe[_TOOL_SEARCH_ENV] = existing
+    if os.environ.get("CLAUDE_CODE_USE_FOUNDRY"):
+        probe["CLAUDE_CODE_USE_FOUNDRY"] = os.environ["CLAUDE_CODE_USE_FOUNDRY"]
     written = _configure_tool_search_env(probe, flag_value)
     return written if written is not None else probe.get(_TOOL_SEARCH_ENV, "")
 
@@ -1478,6 +1480,36 @@ def _write_claude_wrap_base_url(
     return previous
 
 
+def _write_claude_wrap_tool_search(value: str, *, settings_path: Path | None = None) -> str | None:
+    """Persist the resolved tool-search mode for daemon-spawned workers.
+
+    Claude Code workers read project settings afresh rather than inheriting
+    the parent process environment (#2492). Keep this separate from the proxy
+    URL crash marker: a stale tool-search mode cannot route traffic to a dead
+    process, and is restored transactionally when the wrap session exits.
+    """
+    path = settings_path or (Path.cwd() / ".claude" / "settings.local.json")
+    payload = _read_settings_for_write(path)
+    env_map = dict(payload.get("env") or {}) if isinstance(payload.get("env"), dict) else {}
+    previous = env_map.get(_TOOL_SEARCH_ENV)
+    env_map[_TOOL_SEARCH_ENV] = value
+    payload["env"] = env_map
+    path.parent.mkdir(parents=True, exist_ok=True)
+    _write_text(path, json.dumps(payload, indent=2) + "\n")
+    return previous
+
+
+def _restore_claude_wrap_tool_search(
+    previous: str | None, *, settings_path: Path | None = None
+) -> None:
+    """Restore the project-local tool-search value written for this session."""
+    _restore_claude_wrap_base_url(
+        previous,
+        settings_path=settings_path,
+        _key_override=_TOOL_SEARCH_ENV,
+    )
+
+
 def _restore_claude_wrap_base_url(
     previous: str | None,
     *,
@@ -4703,6 +4735,8 @@ def claude(
 
     proxy_holder: list[subprocess.Popen | None] = [None]
     _saved_base_url: list[str | None] = [None]  # previous settings.json value for restore
+    _tool_search_not_written = object()
+    _saved_tool_search: list[object | str | None] = [_tool_search_not_written]
     _settings_foundry: list[bool] = [False]
     port_holder: list[int] = [port]
     _settings_vertex: list[bool] = [False]
@@ -4934,6 +4968,11 @@ def claude(
         # Issue #746: keep Claude Code's on-demand tool loading on through the
         # proxy so tool schemas are not eagerly materialized into local context.
         _tool_search_value = _configure_tool_search_env(env, tool_search)
+        _resolved_tool_search_value = env.get(_TOOL_SEARCH_ENV, "")
+        _saved_tool_search[0] = _write_claude_wrap_tool_search(
+            _resolved_tool_search_value,
+            settings_path=_wrap_settings_path,
+        )
         if _tool_search_value is not None:
             # Describe what the written value actually does: --tool-search
             # false/0/no/off turns deferral OFF, and the banner must say so
@@ -4979,6 +5018,11 @@ def claude(
         click.echo(f"  Error: {e}")
         raise SystemExit(1) from e
     finally:
+        if _saved_tool_search[0] is not _tool_search_not_written:
+            _restore_claude_wrap_tool_search(
+                cast(str | None, _saved_tool_search[0]),
+                settings_path=_wrap_settings_path,
+            )
         _restore_claude_wrap_base_url(
             _saved_base_url[0],
             foundry_mode=_settings_foundry[0],
@@ -5435,8 +5479,8 @@ def vscode_copilot(
 ) -> None:
     """Run Headroom for GitHub Copilot inside Visual Studio Code.
 
-    Transparently overrides Copilot's proxy endpoint, preserving the model
-    selected in VS Code. It does not edit Codex settings.
+    Transparently overrides Copilot's proxy and CAPI endpoints, preserving the
+    model selected in VS Code. It does not edit Codex settings.
     """
     resolution = _require_copilot_subscription_resolution()
     target_settings = settings_file or vscode_settings_path()
@@ -5456,6 +5500,9 @@ def vscode_copilot(
         click.echo(
             f'  "github.copilot.advanced.debug.overrideProxyUrl": "{vscode_proxy_url(actual_port, _project_name_from_cwd())}",'
         )
+        click.echo(
+            f'  "github.copilot.advanced.debug.overrideCapiUrl": "{vscode_proxy_url(actual_port, _project_name_from_cwd())}",'
+        )
         click.echo('  "github.copilot.advanced.debug.overrideAuthType": "token"')
 
     _run_proxy_only_watcher(
diff --git a/headroom/providers/claude/vscode.py b/headroom/providers/claude/vscode.py
index 8f7e62486..79f9f3889 100644
--- a/headroom/providers/claude/vscode.py
+++ b/headroom/providers/claude/vscode.py
@@ -89,7 +89,11 @@ def configure_vscode_claude_settings(path: Path, proxy_url: str) -> str:
     payload = _read_settings(path)
     env = _env_map(payload, path)
     state_path = _state_path(path)
-    managed = {_BASE_URL_KEY: proxy_url, _TOOL_SEARCH_KEY: "true"}
+    # Claude Code's VS Code webview cannot render the server_tool_use /
+    # tool_search_tool_result blocks emitted by deferred tool search (#2028).
+    # Keep it disabled for this surface; the standalone CLI retains its own
+    # configurable/default-on policy.
+    managed = {_BASE_URL_KEY: proxy_url, _TOOL_SEARCH_KEY: "false"}
 
     if state_path.exists():
         state = _read_object(state_path, label="Headroom state")
diff --git a/headroom/providers/copilot/vscode.py b/headroom/providers/copilot/vscode.py
index 2a0fbc85b..354a4efe5 100644
--- a/headroom/providers/copilot/vscode.py
+++ b/headroom/providers/copilot/vscode.py
@@ -17,6 +17,7 @@ from headroom.proxy.project_context import with_project_prefix
 _MARKER_START = "// --- Headroom Copilot proxy ---"
 _MARKER_END = "// --- end Headroom Copilot proxy ---"
 _PROXY_KEY = "github.copilot.advanced.debug.overrideProxyUrl"
+_CAPI_KEY = "github.copilot.advanced.debug.overrideCapiUrl"
 _AUTH_KEY = "github.copilot.advanced.debug.overrideAuthType"
 
 
@@ -118,6 +119,7 @@ def _managed_block(proxy_url: str, *, owns_preceding_comma: bool, line_sep: str)
     return (
         f"\t{marker}{line_sep}"
         f"\t{json.dumps(_PROXY_KEY)}: {json.dumps(proxy_url)},{line_sep}"
+        f"\t{json.dumps(_CAPI_KEY)}: {json.dumps(proxy_url)},{line_sep}"
         f'\t{json.dumps(_AUTH_KEY)}: "token"{line_sep}'
         f"\t{_MARKER_END}"
     )
@@ -161,7 +163,7 @@ def configure_vscode_proxy_settings(path: Path, proxy_url: str) -> str:
     if had_managed_block:
         remove_vscode_proxy_settings(path)
         raw = _read_settings(path)
-    elif _PROXY_KEY in raw or _AUTH_KEY in raw:
+    elif _PROXY_KEY in raw or _CAPI_KEY in raw or _AUTH_KEY in raw:
         raise click.ClickException(
             f"{path} already configures a Copilot endpoint override outside Headroom's "
             "managed block; refusing to replace it. Remove it or use --no-configure."
diff --git a/headroom/providers/openai_responses.py b/headroom/providers/openai_responses.py
index e09adc42d..2e10b030f 100644
--- a/headroom/providers/openai_responses.py
+++ b/headroom/providers/openai_responses.py
@@ -32,9 +32,19 @@ OPENAI_RESPONSES_ROOT_PATHS: tuple[str, ...] = (
     "/v1/codex/responses",
     "/backend-api/responses",
     "/backend-api/codex/responses",
+    # Copilot Chat derives this unprefixed path from overrideCapiUrl. Without an
+    # explicit root route it falls through to uncompressed generic passthrough.
+    "/responses",
 )
 
-OPENAI_RESPONSES_WEBSOCKET_PATHS: tuple[str, ...] = OPENAI_RESPONSES_ROOT_PATHS
+# The Codex websocket relay speaks a different protocol; do not register the
+# Copilot HTTP alias as a websocket route without separate wire validation.
+OPENAI_RESPONSES_WEBSOCKET_PATHS: tuple[str, ...] = (
+    "/v1/responses",
+    "/v1/codex/responses",
+    "/backend-api/responses",
+    "/backend-api/codex/responses",
+)
 
 OPENAI_RESPONSES_SUBPATH_ROUTES: tuple[OpenAIResponsesSubpathRoute, ...] = (
     OpenAIResponsesSubpathRoute("/v1/responses/{sub_path:path}", ("GET", "POST", "DELETE")),
diff --git a/headroom/providers/route_specs.py b/headroom/providers/route_specs.py
index 2aaabc9a4..296269443 100644
--- a/headroom/providers/route_specs.py
+++ b/headroom/providers/route_specs.py
@@ -105,6 +105,9 @@ ANTHROPIC_BATCH_ROUTES: tuple[ProviderHandlerRoute, ...] = (
 
 OPENAI_HANDLER_ROUTES: tuple[ProviderHandlerRoute, ...] = (
     ProviderHandlerRoute("POST", "/v1/chat/completions", "handle_openai_chat"),
+    # Copilot Chat derives this unprefixed path from overrideCapiUrl. Route it
+    # through the real handler; the generic catch-all would bypass compression.
+    ProviderHandlerRoute("POST", "/chat/completions", "handle_openai_chat"),
 )
 
 
diff --git a/tests/test_cli/test_init_enable_tool_search.py b/tests/test_cli/test_init_enable_tool_search.py
index de65e56ed..6606163ae 100644
--- a/tests/test_cli/test_init_enable_tool_search.py
+++ b/tests/test_cli/test_init_enable_tool_search.py
@@ -16,7 +16,8 @@ from headroom.cli import init as init_cli
 from headroom.providers.claude import install as claude_install
 
 
-def test_ensure_claude_hooks_sets_enable_tool_search(tmp_path: Path) -> None:
+def test_ensure_claude_hooks_sets_enable_tool_search(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.delenv("CLAUDE_CODE_USE_FOUNDRY", raising=False)
     settings = tmp_path / "settings.json"
     init_cli._ensure_claude_hooks(settings, profile="init-user", port=8787)
 
@@ -25,6 +26,16 @@ def test_ensure_claude_hooks_sets_enable_tool_search(tmp_path: Path) -> None:
     assert env["ENABLE_TOOL_SEARCH"] == "true"
 
 
+def test_ensure_claude_hooks_disables_tool_search_for_foundry(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("CLAUDE_CODE_USE_FOUNDRY", "1")
+    settings = tmp_path / "settings.json"
+
+    init_cli._ensure_claude_hooks(settings, profile="init-user", port=8787)
+
+    env = json.loads(settings.read_text(encoding="utf-8"))["env"]
+    assert env["ENABLE_TOOL_SEARCH"] == "false"
+
+
 def test_ensure_claude_hooks_respects_user_tool_search_value(tmp_path: Path) -> None:
     settings = tmp_path / "settings.json"
     settings.write_text(
diff --git a/tests/test_cli/test_wrap_claude_base_url.py b/tests/test_cli/test_wrap_claude_base_url.py
index 5c252877e..591ba9c1d 100644
--- a/tests/test_cli/test_wrap_claude_base_url.py
+++ b/tests/test_cli/test_wrap_claude_base_url.py
@@ -34,6 +34,29 @@ def test_write_preserves_other_env_keys(tmp_path: Path) -> None:
     assert payload["env"]["ANTHROPIC_BASE_URL"] == "http://127.0.0.1:8787"
 
 
+def test_tool_search_write_and_restore_reaches_daemon_worker_settings(tmp_path: Path) -> None:
+    path = _settings(tmp_path)
+    path.parent.mkdir(parents=True)
+    path.write_text(
+        json.dumps({"env": {"ENABLE_TOOL_SEARCH": "true", "KEEP": "1"}}),
+        encoding="utf-8",
+    )
+
+    previous = wrap_cli._write_claude_wrap_tool_search("false", settings_path=path)
+
+    assert previous == "true"
+    assert json.loads(path.read_text(encoding="utf-8"))["env"] == {
+        "ENABLE_TOOL_SEARCH": "false",
+        "KEEP": "1",
+    }
+
+    wrap_cli._restore_claude_wrap_tool_search(previous, settings_path=path)
+    assert json.loads(path.read_text(encoding="utf-8"))["env"] == {
+        "ENABLE_TOOL_SEARCH": "true",
+        "KEEP": "1",
+    }
+
+
 def test_write_returns_none_when_key_absent(tmp_path: Path) -> None:
     path = _settings(tmp_path)
     prev = wrap_cli._write_claude_wrap_base_url("http://127.0.0.1:8787", settings_path=path)
diff --git a/tests/test_cli/test_wrap_claude_vertex_proxy_env.py b/tests/test_cli/test_wrap_claude_vertex_proxy_env.py
index 69c8a03f9..e4ec5ea14 100644
--- a/tests/test_cli/test_wrap_claude_vertex_proxy_env.py
+++ b/tests/test_cli/test_wrap_claude_vertex_proxy_env.py
@@ -78,6 +78,13 @@ def _invoke_wrap_claude(
 
     monkeypatch.setattr(wrap_mod, "_write_claude_wrap_base_url", fake_write_base_url)
     monkeypatch.setattr(wrap_mod, "_restore_claude_wrap_base_url", lambda *_args, **_kwargs: None)
+
+    def fake_write_tool_search(value: str, **kwargs: object) -> None:
+        captured["write_tool_search_value"] = value
+        captured["write_tool_search_kwargs"] = kwargs
+
+    monkeypatch.setattr(wrap_mod, "_write_claude_wrap_tool_search", fake_write_tool_search)
+    monkeypatch.setattr(wrap_mod, "_restore_claude_wrap_tool_search", lambda *_a, **_k: None)
     monkeypatch.setattr(wrap_mod, "_print_telemetry_notice", lambda: None)
 
     def fake_ensure_proxy(*args: object, **kwargs: object) -> tuple[None, int]:
@@ -209,6 +216,24 @@ def test_wrap_claude_tool_search_banner_line_still_accurate_when_active(
     assert "on-demand tool loading kept on" in output
     assert "keeps it on for this session" in output
     assert "DISABLED per your setting" not in output
+    assert _captured["write_tool_search_value"] == "true"
+
+
+def test_wrap_claude_foundry_persists_disabled_tool_search_for_workers(
+    runner: CliRunner, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    captured, output = _invoke_wrap_claude(
+        runner,
+        monkeypatch,
+        env={
+            "CLAUDE_CODE_USE_FOUNDRY": "1",
+            "ANTHROPIC_FOUNDRY_BASE_URL": "https://tenant.services.ai.azure.com/anthropic",
+        },
+    )
+
+    assert captured["child_env"]["ENABLE_TOOL_SEARCH"] == "false"
+    assert captured["write_tool_search_value"] == "false"
+    assert "on-demand tool loading DISABLED" in output
 
 
 def test_wrap_claude_vertex_passes_custom_base_url_to_proxy_before_child_redirect(
diff --git a/tests/test_cli/test_wrap_vscode.py b/tests/test_cli/test_wrap_vscode.py
index 6d3496991..fccab5cac 100644
--- a/tests/test_cli/test_wrap_vscode.py
+++ b/tests/test_cli/test_wrap_vscode.py
@@ -66,6 +66,7 @@ def test_wrap_vscode_no_configure_prints_transparent_settings(tmp_path: Path) ->
     assert result.exit_code == 0, result.output
     assert not path.exists()
     assert "overrideProxyUrl" in result.output
+    assert "overrideCapiUrl" in result.output
     assert "overrideAuthType" in result.output
 
 
diff --git a/tests/test_cli/test_wrap_vscode_claude.py b/tests/test_cli/test_wrap_vscode_claude.py
index ec3a37662..e57029240 100644
--- a/tests/test_cli/test_wrap_vscode_claude.py
+++ b/tests/test_cli/test_wrap_vscode_claude.py
@@ -25,7 +25,7 @@ def test_wrap_vscode_claude_configures_actual_port(tmp_path: Path) -> None:
     assert result.exit_code == 0, result.output
     env = json.loads(path.read_text(encoding="utf-8"))["env"]
     assert env["ANTHROPIC_BASE_URL"].startswith("http://127.0.0.1:9999/p/")
-    assert env["ENABLE_TOOL_SEARCH"] == "true"
+    assert env["ENABLE_TOOL_SEARCH"] == "false"
     assert "Reload VS Code" in result.output
     assert captured["agent_type"] == "claude"
 
diff --git a/tests/test_provider_claude_vscode_config.py b/tests/test_provider_claude_vscode_config.py
index 61f9e73de..1e470a3f9 100644
--- a/tests/test_provider_claude_vscode_config.py
+++ b/tests/test_provider_claude_vscode_config.py
@@ -55,7 +55,7 @@ def test_configure_and_remove_preserve_unrelated_and_previous_values(tmp_path: P
     assert configured["env"] == {
         "KEEP": "yes",
         "ANTHROPIC_BASE_URL": "http://127.0.0.1:8787/p/demo",
-        "ENABLE_TOOL_SEARCH": "true",
+        "ENABLE_TOOL_SEARCH": "false",
     }
     assert configured["permissions"] == {"allow": ["Read"]}
 
@@ -162,7 +162,7 @@ def test_reconfigure_refuses_incomplete_or_conflicting_state(tmp_path: Path) ->
     state_path.unlink()
     configure_vscode_claude_settings(path, proxy_url)
     payload = json.loads(path.read_text(encoding="utf-8"))
-    payload["env"]["ENABLE_TOOL_SEARCH"] = "false"
+    payload["env"]["ENABLE_TOOL_SEARCH"] = "true"
     path.write_text(json.dumps(payload), encoding="utf-8")
     with pytest.raises(click.ClickException, match="managed values"):
         configure_vscode_claude_settings(path, proxy_url)
diff --git a/tests/test_provider_copilot_vscode_config.py b/tests/test_provider_copilot_vscode_config.py
index c3075bb83..992888e22 100644
--- a/tests/test_provider_copilot_vscode_config.py
+++ b/tests/test_provider_copilot_vscode_config.py
@@ -50,6 +50,7 @@ def test_configure_update_and_remove_preserve_jsonc_verbatim(tmp_path: Path) ->
     assert "editor.fontSize" in configured
     assert "user comment" in configured
     assert '"github.copilot.advanced.debug.overrideProxyUrl"' in configured
+    assert '"github.copilot.advanced.debug.overrideCapiUrl"' in configured
     assert '"github.copilot.advanced.debug.overrideAuthType": "token"' in configured
 
     assert configure_vscode_proxy_settings(path, "http://127.0.0.1:9999") == "updated"
@@ -66,6 +67,7 @@ def test_configure_refuses_malformed_and_unmanaged_override(tmp_path: Path) -> N
     for original in (
         "{broken",
         '{"github.copilot.advanced.debug.overrideProxyUrl":"http://other"}',
+        '{"github.copilot.advanced.debug.overrideCapiUrl":"http://other"}',
     ):
         path.write_text(original, encoding="utf-8")
         with pytest.raises(click.ClickException, match="did not overwrite|refusing"):
diff --git a/tests/test_provider_openai_responses.py b/tests/test_provider_openai_responses.py
index 439166ae9..45a88a0fa 100644
--- a/tests/test_provider_openai_responses.py
+++ b/tests/test_provider_openai_responses.py
@@ -23,8 +23,15 @@ def test_openai_responses_route_aliases_are_explicit() -> None:
         "/v1/codex/responses",
         "/backend-api/responses",
         "/backend-api/codex/responses",
+        "/responses",
     )
-    assert OPENAI_RESPONSES_WEBSOCKET_PATHS == OPENAI_RESPONSES_ROOT_PATHS
+    assert OPENAI_RESPONSES_WEBSOCKET_PATHS == (
+        "/v1/responses",
+        "/v1/codex/responses",
+        "/backend-api/responses",
+        "/backend-api/codex/responses",
+    )
+    assert "/responses" not in OPENAI_RESPONSES_WEBSOCKET_PATHS
     assert OPENAI_RESPONSES_SUBPATH_ROUTES == (
         OpenAIResponsesSubpathRoute("/v1/responses/{sub_path:path}", ("GET", "POST", "DELETE")),
         OpenAIResponsesSubpathRoute(
diff --git a/tests/test_provider_proxy_routes.py b/tests/test_provider_proxy_routes.py
index 2319e3571..22698365d 100644
--- a/tests/test_provider_proxy_routes.py
+++ b/tests/test_provider_proxy_routes.py
@@ -496,7 +496,9 @@ def test_provider_specific_routes_delegate_to_expected_proxy_handlers(monkeypatc
             "handle_anthropic_batch_passthrough"
         )
         assert client.post("/v1/chat/completions").json()["handler"] == "handle_openai_chat"
+        assert client.post("/chat/completions").json()["handler"] == "handle_openai_chat"
         assert client.post("/v1/responses").json()["handler"] == "handle_openai_responses"
+        assert client.post("/responses").json()["handler"] == "handle_openai_responses"
         assert client.post("/v1/codex/responses").json()["handler"] == "handle_openai_responses"
         assert client.post("/backend-api/responses").json()["handler"] == "handle_openai_responses"
         assert client.post("/backend-api/codex/responses").json()["handler"] == (
diff --git a/tests/test_provider_route_specs.py b/tests/test_provider_route_specs.py
index 52aa8e988..f6a57db0a 100644
--- a/tests/test_provider_route_specs.py
+++ b/tests/test_provider_route_specs.py
@@ -82,6 +82,7 @@ def test_direct_handler_routes_model_endpoint_intent() -> None:
     )
     assert OPENAI_HANDLER_ROUTES == (
         ProviderHandlerRoute("POST", "/v1/chat/completions", "handle_openai_chat"),
+        ProviderHandlerRoute("POST", "/chat/completions", "handle_openai_chat"),
     )
     assert GEMINI_HANDLER_ROUTES == (
         ProviderHandlerRoute(