Filtering Requests
Once you’re on the Requests page, you can search and filter to find a specific request. To zero in on failures, click the filter icon and filter by status - along with a range of other available filters.
Creating Requests
The Requests page isn’t only for observability - click the + button on the right to upload one or more files and select the model and detail level for your request. You can also enable confidence scoring and visual grounding on the request.
Using Skills in Requests
To make a request useful, provide guidance for what you want extracted - the simplest way is to select a skill. The skill handles the prompt, output schema, and plan for the request.
Request Details
Click any request to see the input file alongside the output. The output can be viewed as JSON, Markdown, or Form view - see Executions for details on field-level feedback.
Request Table
| Column | Description |
|---|---|
| Endpoint | The API path called (e.g., /v1/image/generate, /v1/chat/completions) |
| Status | HTTP status code: 200 for success, 4xx/5xx for errors |
| Duration | Wall-clock time from request received to response sent |
| Credits | The number of credits consumed by this request |
| Timestamp | When the request was received |
| Skill / Domain | The skill or domain referenced in the request, if any |
Common Use Cases
- Audit trail: See exactly what was sent and returned for any request, useful for compliance and debugging.
- Cost analysis: Sort by credits to identify the most expensive requests and optimize accordingly.
- Error investigation: Filter for
4xxor5xxstatus codes to find and diagnose failures. - Skill performance: Filter by skill name to compare success rates and latency across skill versions.
Related Pages
Observe Overview
Return to the observability dashboard.
Executions
Track multi-step agent executions.
Generate API
Reference for image, document, audio, and video generate endpoints.
CLI Generate
Run generate requests from the command line.