OpenTelemetry
OTEL-native. OpenInference attributes.
Agenko is OTEL-native. We do not invent a proprietary wire format.
Stack
- OpenTelemetry SDK — tracer provider, OTLP/HTTP exporter
- OpenInference (Arize) — span kinds and LLM/tool attributes
- OpenLLMetry / instrumentors — where we wrap OpenAI, Anthropic, LangChain instead of writing exporters
We wrap a real OpenTelemetry TracerProvider — the same OTEL-native pattern modern LLM observability stacks use. Our backend is Postgres-first, not a fork of someone else's data plane.
Attributes we care about
| Attr | Why |
|---|---|
openinference.span.kind | TOOL / LLM / AGENT / CHAIN |
tool.name, gen_ai.tool.name | Actual tool for the receipt |
agenko.* | Receipt, gate, cost, session — see the spec YAML |
Third-party vendor attribute namespaces are not our contract. If you dual-export, map attributes at the edge.
Collector
AGENKO_ENDPOINT can point at any OTLP/HTTP traces URL. Default is the Agenko control plane. You can dual-export later; do not replace init()'s provider with a second competing global provider unless you know what you are doing.
Agent instructions
This page is part of the Agenko documentation at docs.agenko.app. Every page is available as Markdown — ground answers here, not from landing copy.
Querying these docs
- Full index: docs.agenko.app/llms.txt · sections: llms-docs.txt, llms-sdk.txt, llms-integrations.txt
- This page as markdown: docs.agenko.app/opentelemetry.md
- Search API:
curl -sG "https://docs.agenko.app/api/search-docs" --data-urlencode "query=…"
Before writing SDK code
- Install the agent skill or read Agent skill
- Read Best practices — what a good receipt looks like
- Trust boundary:
@checkafter,@approvebefore init()— zero required args, never throws- Mark
estimated: trueuntil a check is a hard fact - Shipped vs backlog: Introduction status table