Switch to AsyncAnthropic client (#12)

Fixes #9

The sync `Anthropic` client was being used inside `async` FastAPI
routes, blocking the event loop during API calls. As @Alc-Alc correctly
identified, this would affect throughput under concurrent requests.

### Changes
- `app/explain.py`: `Anthropic` → `AsyncAnthropic`, `await
client.messages.create()`
- `app/main.py`: `Anthropic()` → `AsyncAnthropic()` in lifespan
- `app/test_explain.py`: `MagicMock` → `AsyncMock` for the
messages.create mock

All 91 tests pass, live smoke test confirmed working.

*(I'm Molty, an AI assistant acting on behalf of @mattgodbolt)*

Author: mattgodbolt

app/explain.py

@@ -1,6 +1,6 @@
 import logging
 
-from anthropic import Anthropic
+from anthropic import AsyncAnthropic
 
 from app.cache import CacheProvider, cache_response, get_cached_response
 from app.explain_api import CostBreakdown, ExplainRequest, ExplainResponse, TokenUsage
@@ -19,7 +19,7 @@
 
 async def process_request(
     body: ExplainRequest,
-    client: Anthropic,
+    client: AsyncAnthropic,
     prompt: Prompt,
     metrics_provider: MetricsProvider,
     cache_provider: CacheProvider | None = None,
@@ -31,7 +31,7 @@ async def process_request(
 
     Args:
         body: The request body as a Pydantic model
-        client: Anthropic client instance
+        client: AsyncAnthropic client instance
         prompt: Prompt instance for generating messages
         metrics_provider: metrics provider for tracking stats
         cache_provider: cache provider for storing/retrieving responses
@@ -69,7 +69,7 @@ async def process_request(
 
 async def _call_anthropic_api(
     body: ExplainRequest,
-    client: Anthropic,
+    client: AsyncAnthropic,
     prompt: Prompt,
     metrics_provider: MetricsProvider,
 ) -> ExplainResponse:
@@ -92,7 +92,7 @@ async def _call_anthropic_api(
     # Call Claude API
     LOGGER.info("Using Anthropic client with model: %s", {prompt_data["model"]})
 
-    message = client.messages.create(
+    message = await client.messages.create(
         model=prompt_data["model"],
         max_tokens=prompt_data["max_tokens"],
         temperature=prompt_data["temperature"],

app/main.py

@@ -2,7 +2,7 @@
 from contextlib import asynccontextmanager
 from pathlib import Path
 
-from anthropic import Anthropic
+from anthropic import AsyncAnthropic
 from anthropic import __version__ as anthropic_version
 from fastapi import FastAPI, Request
 from fastapi.middleware.cors import CORSMiddleware
@@ -42,7 +42,7 @@ async def lifespan(app: FastAPI):
 
     # Store shared resources in app.state
     app.state.settings = settings
-    app.state.anthropic_client = Anthropic(api_key=settings.anthropic_api_key)
+    app.state.anthropic_client = AsyncAnthropic(api_key=settings.anthropic_api_key)
 
     # Load the prompt configuration
     prompt_config_path = Path(__file__).parent / "prompt.yaml"

app/test_explain.py

@@ -1,6 +1,6 @@
 import json
 from pathlib import Path
-from unittest.mock import MagicMock
+from unittest.mock import AsyncMock, MagicMock
 
 import pytest
 
@@ -65,7 +65,7 @@ def mock_anthropic_client():
     mock_message.usage.input_tokens = 100
     mock_message.usage.output_tokens = 50
 
-    mock_client.messages.create.return_value = mock_message
+    mock_client.messages.create = AsyncMock(return_value=mock_message)
     return mock_client