Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
Python
import asyncio
import threading
from unittest.mock import AsyncMock, patch
import httpx
import pytest
from anyio import to_thread
from free_claude_code.config.loader import ManagedConfigStore
from tests.api.support import create_test_app
@pytest.mark.asyncio
@pytest.mark.parametrize(
"path", ["config", "status", "providers/lmstudio/local-status", "config/apply"]
)
async def test_admin_storage_wait_allows_health_and_authenticated_requests(path):
app = create_test_app()
entered, release = threading.Event(), threading.Event()
read = ManagedConfigStore.read
worker_limiter = to_thread.current_default_thread_limiter()
original_tokens = worker_limiter.total_tokens
def blocked_read(store, *args):
entered.set()
assert release.wait(5), "Admin blocked the event loop"
return read(store, *args)
async with httpx.AsyncClient(
transport=httpx.ASGITransport(app=app, client=("127.0.0.1", 50000)),
base_url="http://127.0.0.1",
) as client:
async def health_during_read():
while not entered.is_set():
await asyncio.sleep(0)
try:
assert (await client.get("/health")).status_code == 200
response = await asyncio.wait_for(client.get("/"), timeout=1)
assert response.status_code == 200
finally:
release.set()
observer = asyncio.create_task(health_during_read())
try:
# Model the last shared worker being occupied by configuration I/O.
worker_limiter.total_tokens = 1
with (
patch.object(ManagedConfigStore, "read", blocked_read),
patch(
"free_claude_code.api.admin_routes._check_local_provider",
AsyncMock(return_value={}),
),
):
if path == "config/apply":
response = await client.post(
f"/admin/api/{path}", json={"values": {"PORT": None}}
)
else:
response = await client.get(f"/admin/api/{path}")
assert response.status_code == 200
await asyncio.wait_for(observer, 5)
finally:
release.set()
observer.cancel()
worker_limiter.total_tokens = original_tokens