Tracelet instrument

Skill jnMetaCode/skillet/skills/tracelet-instrument

skillet — a package manager for AI agent skills (SKILL.md). Find, install, version & share skills from a Git-backed registry. Zero infra, MCP-native, reproducible.

Install

npx -y skills add jnMetaCode/skillet --skill tracelet-instrument

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Instrument an AI agent or LLM app with OpenTelemetry and watch its runs live in tracelet (local DevTools). Use when asked to add tracing/observability to an agent, or to debug what an agent actually did.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.6 KB, as published. Nobody here has run it

tracelet-instrument

Goal: make an agent's execution visible — every LLM call, tool call, prompt, token count and latency — without sending anything to a cloud service. tracelet is a local OTLP collector + UI; any OpenTelemetry exporter pointed at http://127.0.0.1:4318 shows up live.

Procedure

Start the collector (keep it running in a terminal):

npx @jnmetacode/tracelet        # ingests OTLP on :4318, UI on :4321

Point the app's exporter at it. For most OpenTelemetry SDKs the zero-code route is environment variables:
```
export OTEL_EXPORTER_OTLP_ENDPOINT=http://127.0.0.1:4318
export OTEL_SERVICE_NAME=my-agent
```

Framework-specific wiring (pick the one that matches the codebase):

Vercel AI SDK — pass telemetry on each call and register a Node tracer:

const result = await generateText({
  model, prompt,
  experimental_telemetry: { isEnabled: true },
});

OpenTelemetry JS (manual spans):

import { NodeSDK } from '@opentelemetry/sdk-node';
import { OTLPTraceExporter } from '@opentelemetry/exporter-trace-otlp-http';
new NodeSDK({ traceExporter: new OTLPTraceExporter({ url: 'http://127.0.0.1:4318/v1/traces' }) }).start();

Python: pip install opentelemetry-sdk opentelemetry-exporter-otlp-proto-http, then the same OTEL_EXPORTER_OTLP_ENDPOINT env var works.

Name spans so the waterfall reads like the agent's plan: one root span per run (agent.run), one child per tool call (tool.<name>) and per model call. Attach gen_ai.request.model, input.value, output.value attributes — tracelet recognizes the GenAI/OpenInference conventions and renders prompts, tokens and tool I/O.
Verify: trigger one agent run, open http://127.0.0.1:4321, and confirm the trace shows the expected span tree with prompts/tokens. If nothing arrives, the exporter is usually pointed at the wrong port (4318) or batching hasn't flushed — force a flush/shutdown on process exit.

Both OTLP/HTTP protobuf (SDK default) and JSON are accepted — no exporter config gymnastics needed.

One thing to look at

What its author says it does

SKILL.md

tracelet-instrument

Procedure

Keep looking