Debug httpx with Smello

httpx is the modern Python HTTP client with first-class async support. It's the transport layer behind the OpenAI, Anthropic, and many other SDKs. Smello patches both httpx.Client (sync) and httpx.AsyncClient (async) automatically.

Setup

pip install smello smello-server
smello-server  # start the dashboard

Then run your script with smello run:

smello run my_app.py

Both sync and async httpx clients are patched. No code changes needed.

Example scripts: basic_httpx.py, async_httpx.py

Scenario: debugging a streaming response that stops mid-stream

You're using httpx to call an API that returns a streaming response, but the stream seems to cut off early. Is the server closing the connection? Is there an error mid-stream?

async with httpx.AsyncClient() as client:
    resp = await client.post(
        "https://api.example.com/v1/generate",
        json={"prompt": "Write a poem", "stream": True},
    )
    # Response status is 200, but the body looks truncated

Debug in the dashboard

Open the Smello dashboard. The captured response shows:

• Status code: confirms the server returned 200 (not a 5xx mid-stream).
• Response headers: check Content-Type, Transfer-Encoding, and any API-specific headers like rate limit counters.
• Response body: the full body as received by httpx, so you can see exactly where it ended.
• Duration: a suspiciously short duration might indicate the connection was dropped rather than the response being complete.
• Remote host group: wraps adjacent loaded requests by destination inside their runtime parent operations.

Debug with an AI agent

If you use Claude Code or another AI coding tool, the /smello skill can query captured events and cross-reference them with your source code. Install it once:

npx skills add smelloscope/smello --skill smello

Then ask your agent:

/smello
My streaming response from api.example.com is getting truncated

The skill is also invoked automatically when your agent recognizes a debugging question, but calling /smello explicitly gives the best results. See AI Agent Skills for compatible tools.

Tips

• Async and sync: Smello patches both httpx.Client.send and httpx.AsyncClient.send, so you don't need separate configuration.
• Timeout debugging: If a request times out, Smello records a failed operation with the exception type and message, then HTTPX raises the original timeout.
• Streaming context: Smello retains the operation context until a streaming response closes, so delayed body capture keeps the original trace, parent span, and memberships.
• SDK traffic: If you're debugging OpenAI or Anthropic SDK calls, those use httpx under the hood. You'll see the raw HTTP traffic in Smello even though you're calling a high-level SDK.
• HTTP/2: httpx supports HTTP/2, and Smello captures these requests the same way. The library field in the dashboard shows httpx.

Next step: follow the getting started guide for installation, configuration, and more.