Gives Claude the ability to instrument your LLM agents with Opik tracing and observability. It handles both the decorator setup and the entrypoint marking needed for Local Runner integration, follows your existing config style instead of forcing new patterns, and knows the integrations for OpenAI, Anthropic, LangChain, and others. The thread grouping for multi-turn conversations is genuinely useful if you're building chat agents. One gotcha it handles well: the LiteLLM current_span_data dance that prevents orphaned traces when you mix manual tracking with callbacks. Includes prompt versioning and eval suite setup, though those are secondary to the core instrumentation work.
- skills.sh installs
- 548
- GitHub stars
- 8
- last updated
- Aug 21, 2026
Install
$ npx -y skills add comet-ml/opik-skills --skill opik --agent claude-codeInstalls into .claude/skills of the current project.
Files
Opik SDK Reference
Opik is an open-source LLM observability platform. This skill is a reference
for the SDK. To instrument a codebase step by step (detect frameworks, add
config, emit and verify a trace), use the task-shaped opik-instrument skill.
Core concepts
A trace is one execution path (one request → one response). Spans are the operations inside it and form a hierarchy.
Span types — the ONLY valid values
| Type | Use for |
|---|---|
general | orchestration, agent entry points |
llm | model calls |
tool | tools, retrieval, API / DB calls |
guardrail | safety / validation checks |
Do NOT use retrieval or any other value.
Python — tracing
import opik
@opik.track(name="agent", type="general")
def agent(query: str) -> str:
return generate(retrieve(query))
@opik.track(type="tool")
def retrieve(query): ...
@opik.track(type="llm")
def generate(ctx): ...
opik.flush_tracker() # required in scripts
TypeScript — tracing
import { Opik } from "opik";
const client = new Opik({ projectName: "my-project" });
Sponsors
Details
- Repository
- github.com/comet-ml/opik-skills
- Categories
- Testing & QA, DevOps & CI/CD, AI & Agent Building, Data Science & ML
- First seen
- Jun 3, 2026
