Python SDK for Pydantic Logfire, an observability tool that instruments Python apps and LLM agents with OpenTelemetry traces, metrics and logs.
comet-ml/opik-openclaw
Official OpenClaw plugin that exports agent traces, tool calls, token usage and costs to Opik for observability.
- Stars
- 726
- Forks
- 87
- Open issues
- 17
- Last commit
- 17 days ago
What is it?
opik-openclaw is an official plugin for OpenClaw that sends agent run data to Opik, the open-source LLM observability and evaluation platform. It records LLM request and response spans, sub-agent spans, tool call spans with inputs, outputs and errors, and run-level metadata including usage and cost.
The plugin runs inside the OpenClaw Gateway process, so on a remote gateway it must be installed and configured on that host. Setup is done through an interactive CLI wizard (openclaw opik configure), and settings can also come from environment variables such as OPIK_API_KEY and OPIK_PROJECT_NAME.
It maps OpenClaw lifecycle events such as llm_input, llm_output, before_tool_call, after_tool_call, subagent_spawned and agent_end onto Opik traces and spans, closing any pending spans when a run ends.
Key features
- 01Traces LLM requests and responses as Opik spans.
- 02Captures tool call inputs, outputs, errors and duration.
- 03Records sub-agent lifecycle spans on the requester trace.
- 04Reports token usage and cost metadata for each run.
- 05Interactive setup wizard and status command for checking configuration.
Use cases
- Monitoring cost and token consumption of OpenClaw agents.
- Debugging failed tool calls in multi-step agent runs.
- Inspecting sub-agent behaviour across a conversation trace.
Quick start
- 1
Install the plugin in OpenClaw
openclaw plugins install clawhub:@opik/opik-openclaw - 2
Configure the plugin with the setup wizard
openclaw opik configure - 3
Start the gateway and send a test message
openclaw gateway run
Things to know
- Requires OpenClaw 2026.9.7 or newer, Node.js 24.16 or newer, and npm 10 or newer.
- Needs an Opik account or self-hosted Opik instance and an API key for configuration.
- Must be installed on the host running the OpenClaw Gateway, which matters for remote gateways.
In collections
Topics
FAQ
- What is opik-openclaw?
- Official OpenClaw plugin that exports agent traces, tool calls, token usage and costs to Opik for observability. opik-openclaw is an official plugin for OpenClaw that sends agent run data to Opik, the open-source LLM observability and evaluation platform. It records LLM request and response spans, sub-agent spans, tool call spans with inputs, outputs and errors, and run-level metadata including usage and cost.
- Is opik-openclaw free to use? What license does it have?
- opik-openclaw is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is opik-openclaw still maintained?
- Actively maintained. The last commit to opik-openclaw was on Oct 5, 2026.
- What can I use opik-openclaw for?
- Monitoring cost and token consumption of OpenClaw agents. Debugging failed tool calls in multi-step agent runs. Inspecting sub-agent behaviour across a conversation trace.
- How do I install opik-openclaw?
- Install the plugin in OpenClaw: openclaw plugins install clawhub:@opik/opik-openclaw Configure the plugin with the setup wizard: openclaw opik configure Start the gateway and send a test message: openclaw gateway run
- Where can I get opik-openclaw?
- The source code and installation instructions are on GitHub: https://github.com/comet-ml/opik-openclaw
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source LLM observability and evaluation platform for tracing AI agents, scoring LLM outputs, managing prompts, and monitoring production.
Open-source observability platform for searching and visualizing logs, traces, metrics, session replays and errors on top of ClickHouse.
A vendor-agnostic service that receives, processes and exports traces, metrics and logs to multiple observability backends.
Open-source OpenTelemetry-based observability extensions that trace LLM calls, vector DB queries and agent frameworks in Python apps.
Datadog's open-source agent that collects metrics, logs, traces and profiles from hosts and sends them to the Datadog platform.