A telemetry layer for libraries using
httpxunderneath with a pluggable query user interface.
Caution
This project is under heavy development and is not ready for production use.
Over the course of 2025 our team has been working on several projects that use httpx underneath. Examples of these are our gallagher (which is more formally maintained) but others where we have been intergrating to property management systems where we make calls to their APIs via a custom designed middleware. Our general approach here is to write thin wrappers over their APIs (most of them don't provide an SDKs) which inturn would use httpx.
The solutions include self healing mechanisms, rate limiting, caching and retries. What is generally missing is telemetry of the requests made, specially in cases where the API does not provide sufficient logging, in particular when things go wrong (not to mention that some APIs don't always respond with appropriate error codes which makes logging to a standard service like Logfire less useful).
With a shift back to server side applications, we at Anomaly have been using Django for most of our projects. This project aims to build what initially will be a very simple telemetry layer for httpx requests made by libraries used in Django applications. To start we will simply write to PostgreSQL database tables using table partioning for performance.
Additionally we want to provide a set of user interfaces that the implementing application can provide inline as a query tool for these requests. The typical use case for the visibility is:
- Inspecting the response of the failed requests
- Getting an oversight of the HTTP headers
- Inspect the payload, most of which will be dyanmically generated JSON data.
The inspiration of the user interface is very much what Stripe's Log feature in their Workbench provides, allow the user to:
- Filter requests by status codes, endpoints, time ranges
- Search for specific terms in the request/response payloads
- View request/response headers in a structured way
- Paginate through large sets of requests
Distributed under the MIT License except Artwork and Branding assets.