
Coding Harness Tracing
Ship Claude Code agent work with automatic OpenInference session traces into Arize AX or Phoenix for debugging and production visibility.
Overview
coding-harness-tracing is a plugin marketplace for the Operate phase that automatically traces Claude Code sessions to Arize AX or Phoenix using OpenInference spans.
What is this marketplace?
- 1-plugin marketplace: claude-code-tracing (MIT v1.0.1)
- Automatic tracing of Claude Code sessions with OpenInference spans
- Export to Arize AX or Phoenix for agent observability
- Category: observability with tracing, agents, OpenInference, Arize, Phoenix keywords
- 1 plugin in marketplace
- Plugin version 1.0.1 MIT
- Targets Arize AX and Phoenix backends
Community signal: 12 GitHub stars.
What problem does it solve?
Agent coding sessions are opaque when something breaks in production and you have no standardized trace to replay or compare.
Who is it for?
Indie builders and small teams already on Claude Code who want Phoenix or Arize AX visibility without building a custom tracing pipeline.
Skip if: Builders who only need local one-off chats with no persistence, or teams not using Arize/Phoenix backends.
What do I get? / Deliverables
After install, Claude Code sessions produce exportable OpenInference traces in Arize or Phoenix so you can debug and monitor harness behavior with evidence.
- Automatic OpenInference spans for Claude Code sessions
- Queryable traces in Arize AX or Phoenix
- One-plugin observability bundle (v1.0.1)
Plugins in this marketplace
1 plugin — install individually after you add the marketplace.
Recommended Marketplaces
Journey fit
Observability and tracing belong on the Operate shelf because they matter after agents are in daily use, not during initial ideation. Monitoring is the canonical subphase for harness-level trace export, span inspection, and session-level agent telemetry.
How it compares
Observability plugin marketplace for Claude Code tracing, not an MCP server or prompt skill.
Common Questions / FAQ
Who is Coding Harness Tracing for?
Developers and operators using Claude Code who need session-level traces in Arize AX or Phoenix for agent debugging and monitoring.
When should I use Coding Harness Tracing?
Use it once Claude Code is part of your daily ship loop and you need durable OpenInference spans instead of console-only logs.
How do I add Coding Harness Tracing to my agent?
Register the Arize AI marketplace in Claude Code, install the claude-code-tracing plugin from the bundle, and configure your Arize or Phoenix project per the repo README.