🦜 Beak - LLM Trace Collection

API Endpoint

POST /api/traces/[collection_name]/[flow_name]

Request Format

Send a multipart/form-data request with the following fields:

• inputs - JSON string containing input data
• outputs - JSON string containing output data
• configuration - JSON string containing configuration data
• files - (Optional) One or more files to attach to the trace

Example using cURL

curl -X POST http://localhost:3000/api/traces/my-collection/my-flow \
  -F 'inputs={"prompt":"Hello, world!","model":"gpt-4"}' \
  -F 'outputs={"response":"Hi there!","tokens":15}' \
  -F 'configuration={"temperature":0.7,"max_tokens":100}' \
  -F 'files=@/path/to/file1.pdf' \
  -F 'files=@/path/to/file2.png'

Response

{
  "success": true,
  "traceId": "550e8400-e29b-41d4-a716-446655440000",
  "collection": "my-collection",
  "flow": "my-flow",
  "files": ["550e8400-e29b-41d4-a716-446655440000.trace.0001.pdf"],
  "storagePath": "traces/my-collection/my-flow/550e8400-e29b-41d4-a716-446655440000.trace.json"
}

Test the API

Use the form below to test the API endpoint:

View Traces

Browse all traces in the database with advanced filtering and sorting: