Test plugin realtime inference

Test a plugin’s realtime inference endpoint and return the raw response.

Use this to debug custom plugin inference — especially when retrievers fail with “Cannot extract dense vector” errors. This endpoint calls the same inference path that retrievers use, but returns the raw response instead of trying to extract an embedding.

What to check in the response:

• response_type: Should be “dict” for embedding plugins
• response_keys: Should contain “embedding” or similar key
• raw_response: The actual data your plugin returns

Example:

POST /{plugin_id}/realtime/test
{"inputs": {"text": "hello world"}}

POST /v1/namespaces/{namespace_id}/plugins/{plugin_id}/realtime/test

Test plugin realtime inference

cURL

curl --request POST \
  --url https://api.mixpeek.com/v1/namespaces/{namespace_id}/plugins/{plugin_id}/realtime/test \
  --header 'Content-Type: application/json' \
  --data '
{
  "inputs": {},
  "parameters": {}
}
'

200

success

{
  "status": "success",
  "inference_name": "custom_plugin_my_embedder_1_0_0",
  "raw_response": {
    "embedding": [
      0.1,
      0.2,
      0.3
    ]
  },
  "response_type": "dict",
  "response_keys": [
    "embedding"
  ],
  "message": "Inference call to 'custom_plugin_my_embedder_1_0_0' succeeded"
}

Headers

Authorization · string

REQUIRED: Bearer token authentication using your API key. Format: 'Bearer sk_xxxxxxxxxxxxx'. You can create API keys in the Mixpeek dashboard under Organization Settings.

Examples:

"Bearer YOUR_API_KEY"

"Bearer YOUR_STRIPE_API_KEY"

Path Parameters

namespace_id · string · required

plugin_id · string · required

Body

application/json

Request to test a plugin's realtime inference endpoint.

Use this to debug plugin inference responses before using them in retrievers.

inputs · Inputs · object

Input data to pass to the inference service (e.g., {'text': 'hello'})

parameters · Parameters · object

Additional parameters for the inference call

Response

200

application/json

Inference test result

Response from testing a plugin's realtime inference endpoint.

Returns the raw inference response so users can verify their plugin is returning the expected format (e.g., embedding vectors).

status · string · required

Test result status: 'success' or 'error'

inference_name · string · required

The resolved inference service name that was called

raw_response · any

Raw response data from the inference service

response_type · string | null

Python type of the response (e.g., 'dict', 'list')

response_keys · string[] | null

Top-level keys if response is a dict

error · string | null

Error message if test failed

message · string | null

Human-readable status message