@tangle-network/sdk-telemetry
v0.1.21
Published
Usage tracking, credit management, execution-trace collection, and telemetry export for the Agent Dev Container SDK.
Readme
@tangle-network/sdk-telemetry
Usage tracking, credit management, execution-trace collection, and telemetry export for the Agent Dev Container SDK.
LLM call capture
Capture LLM calls (model, tokens, latency, prompt/completion) with
OpenLLMetry, which auto-instruments
the raw OpenAI/Anthropic SDKs plus 10+ frameworks and emits standard OpenTelemetry
gen_ai.* spans.
- Instrument your app with OpenLLMetry (
traceloop-sdkfor Python,@traceloop/node-server-sdkfor Node). - Point its OTLP exporter at our trace ingest:
POST /v1/otlp/v1/traces(the intelligence OTLP/HTTP-JSON adapter).
Our ingest reads the standard gen_ai.* attribute keys directly off the span
attribute bag — model from gen_ai.request.model, tokens from
gen_ai.usage.prompt_tokens / gen_ai.usage.completion_tokens (and the
input_tokens / output_tokens spellings). The reader candidate arrays live in
packages/sdk-core/src/telemetry/genai-attributes.ts; the ingest reader is
parseGenAiAttribution (products/intelligence/api/src/lib/trace-attribution.ts).
No bespoke provider wrapper is required.
