Python
Set up Python
Skip to content
What WildEdge does
See where model behavior changes across releases, runtimes, and devices. Inspect captured inputs, outputs, and feedback, then turn selected examples into datasets for evaluation and training.
WildEdge instruments your inference calls and records what actually happened in production: latency, confidence, drift, hardware and thermal state, and the feedback your users give on each result. Turn on capture and the raw inputs and outputs land next to the event that produced them.
- Compare model behavior across app releases, model versions, runtimes, and device classes
- Inspect the events that went wrong, with their captured inputs, outputs, and user feedback
- Build datasets from selected examples and export them to your evaluation and training pipelines
Capture is opt-in. Leave it off and the SDK sends telemetry only, never raw inputs or outputs.
Investigate with an AI agent
Connect Claude Code, Codex, Gemini CLI, or ChatGPT to your account through the remote MCP server. Ask it to inspect recent inference events, follow a trace, or dig into a failure, without copying data between your browser and your terminal.
- Connect a coding agent with a personal access token
- Connect a web agent such as ChatGPT or Claude with OAuth
Examples
Every SDK ships runnable examples: Python, Swift, Android.
The demo app runs on-device and remote models in one pipeline, so you can watch the whole loop before wiring up your own.
API Reference
Interactive Swagger docs are also available at app.wildedge.dev/api/docs.