Trace coding-agent and LLM workflows with OpenLIT
Use OpenLIT to instrument coding agents and LLM applications with OpenTelemetry traces, metrics, costs, prompts, tool calls, and evaluation signals.
npx skills add agentskillexchange/skills --skill trace-coding-agent-and-llm-workflows-with-openlit
git clone https://github.com/openlit/openlit.git and docker compose up -d; install the Python SDK with pip install openlit; for coding-agent hooks run the OpenLIT CLI installer, then openlit configure --endpoint http://127.0.0.1:4318 and openlit coding install --vendor=all.OpenLIT is an OpenTelemetry-native AI engineering platform with SDKs, a CLI, dashboards, evaluations, prompt management, guardrails, and integrations across LLM providers, frameworks, vector databases, and GPUs.
What this skill actually does
Use this skill when an operator needs repeatable observability for local coding agents or LLM applications: install the OpenLIT stack, wire SDK or CLI instrumentation into Claude Code, Cursor, Codex, or application code, send traces to an OTLP endpoint, and inspect sessions, prompts, costs, tool calls, file edits, exceptions, and evaluation signals before changing production workflows.
The boundary is trace-backed AI workflow inspection and release review. It is not a generic observability platform card because the invocation focuses on OpenLIT’s LLM/coding-agent instrumentation, `openlit.init()`, OTLP export, and `openlit coding install` workflow.
Source evidence: upstream documents one-line OpenTelemetry-native observability, Python/TypeScript/Go SDKs, 11 built-in evaluation types, and a CLI that installs vendor hooks for Claude Code, Cursor, and Codex sessions.
Inputs and prerequisites: OpenLIT stack, OpenLIT SDK or CLI, OTLP endpoint or local OpenLIT dashboard.
Setup notes: Self-host with `git clone https://github.com/openlit/openlit.git` and `docker compose up -d`; install the Python SDK with `pip install openlit`; for coding-agent hooks run the OpenLIT CLI installer, then `openlit configure –endpoint http://127.0.0.1:4318` and `openlit coding install –vendor=all`.
Source and verification boundary: use https://docs.openlit.io as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.