respan-instrumentation-watsonx 0.1.0

Respan instrumentation plugin for IBM watsonx.ai

Respan IBM watsonx.ai Instrumentation

Native Respan instrumentation for the official ibm-watsonx-ai Python SDK.

Installation

pip install respan-ai respan-instrumentation-watsonx ibm-watsonx-ai

Usage

from respan import Respan
from respan_instrumentation_watsonx import WatsonxInstrumentor

respan = Respan(
    api_key="RESPAN_API_KEY",
    instrumentations=[WatsonxInstrumentor()],
)

The instrumentor traces:

• ModelInference.generate()
• ModelInference.generate_text()
• ModelInference.generate_text_stream()
• ModelInference.chat()
• ModelInference.chat_stream()
• ModelInference.agenerate()
• ModelInference.agenerate_stream()
• ModelInference.achat()
• ModelInference.achat_stream()
• Embeddings.generate()
• Embeddings.embed_query()
• Embeddings.embed_documents()
• Embeddings.agenerate()
• Embeddings.aembed_query()
• Embeddings.aembed_documents()

Chat and text-generation spans use canonical GenAI fields such as gen_ai.prompt.*, gen_ai.completion.*, llm.request.functions, and token usage attributes. Embedding spans record the input and usage metadata but do not export embedding vectors.