Full trace history
Every session is recorded as a trace, including model turns, tool calls, and spans, so you can inspect a full run rather than only a summary number.
LangWatch
LangWatch tracks Claude Code sessions as traces with model turns, tool calls, token classes, and cost data. It also supports Codex, OpenCode, Copilot, Cursor, and Pi through the same OTLP workflow.
Track Claude Code usage with trace-level accounting
LangWatch is a trace and usage tracking product for Claude Code and related coding agents. It captures each session as a trace, including model turns, tool calls, token counts by class, and cost data, so developers can review what happened in a run and how usage accumulated.
The Claude Code page emphasizes full trace history and token accounting across bundled subscription plans, where LangWatch can display theoretical spend rather than a direct token bill. It also extends to Codex, OpenCode, Copilot, Cursor, and Pi through the same OTLP endpoint, which makes it useful for comparing agent usage in a single view.
What LangWatch tracks for Claude Code
Token classes by span
Token accounting breaks out cache reads and cache writes separately from input and output tokens, giving a more detailed view of where usage accumulates.
Theoretical spend for bundled plans
The page shows both billed status and a theoretical total, which is useful for understanding spending on bundled subscription plans that are not invoiced per token.
OTel-based setup
Native OpenTelemetry export lets Claude Code send usage into LangWatch through `npx langwatch claude`, instead of relying only on local logs.
Multi-agent support
The same endpoint is described as working for Codex, OpenCode, Copilot, Cursor, and Pi, making it possible to compare agents in one store.
Agent-accessible traces
LangWatch MCP can expose trace history back to the agent itself, so Claude Code can inspect past runs and use them while adjusting workflow.
When LangWatch is useful
Post-session trace review
Use LangWatch to review a Claude Code session after it finishes, including model turns, tool calls, token classes, spans, and total duration, so you can understand where time and usage went.
Cost analysis for bundled plans
Use the trace history and token breakdown to understand what bundled-plan sessions would have cost on a per-token basis, especially when you want a theoretical spend view instead of invoice data.
Cross-agent usage comparison
Use the shared OTLP endpoint to compare usage patterns across Claude Code, Codex, OpenCode, Copilot, Cursor, and Pi in one store, rather than switching between separate tools.
Agent workflow inspection
Use LangWatch MCP when you want the agent itself to inspect earlier runs, find where tokens are going, and adapt its workflow from trace history.
Pros and Cons
Pros
- Captures full trace history, not just a single usage total.
- Breaks out token classes, including cache reads and cache writes.
- Shows theoretical spend for bundled plans where there is no per-token invoice.
- Works across multiple coding agents through the same endpoint.
- Can route trace history back to the agent through LangWatch MCP.
Cons
- The page does not fully document setup steps beyond the `npx langwatch claude` command.
- Coverage details such as retention, self-hosting differences, and exact tracked outputs are not fully specified on the Claude Code page itself.
FAQ
How do I track Claude Code usage with LangWatch?
Run `npx langwatch claude` to configure Claude Code’s native OpenTelemetry export to LangWatch. The page says each session then lands as a trace with model turns, tool calls, token classes, and theoretical cost.
Does it work with other coding agents besides Claude Code?
Yes. The page says the same OTLP endpoint works for Codex, OpenCode, Copilot, Cursor, and Pi, so those agents can land in the same store with the same token, cost, and trace treatment.
What does the Bundled cost label mean?
The page describes sessions on Claude subscription plans such as Pro, Max, and Team as “Bundled.” In that case, LangWatch shows the theoretical total spend rather than a per-token invoice amount.
Is there a free plan?
The pricing page shows a free Developer plan, paid Growth pricing from €29 per core seat per month, and an Enterprise plan with custom pricing. It also states you can start free with no credit card.
Why use LangWatch instead of a local usage log reader?
The page indicates the OpenTelemetry route follows you across machines and survives local log cleanup, unlike local log readers that only see one machine.
Quick Facts
- Category
- Developer Tool
- Primary use
- Claude Code usage tracking and trace analysis
- Supported agents
- Claude Code, Codex, OpenCode, Copilot, Cursor, Pi
- Setup shown
- `npx langwatch claude`
- Pricing signal
- Free Developer plan available; paid plans from €29/core-seat/month
- Website
- langwatch.ai
Alternativas ao LangWatch
PromptScout
PromptScout tracks how ChatGPT, Gemini, Google AI Overviews, and Perplexity mention your brand or competitors, then pairs those results with source analysis and website audits. It helps teams decide what to fix in content, positioning, or site readiness next.
Sleek Analytics
Sleek Analytics is a privacy-friendly web analytics tool with real-time visitor tracking, Core Web Vitals, and revenue attribution. It helps site owners understand traffic and conversions without cookie banners or a heavy setup.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
SaveMRR
SaveMRR is a Stripe retention tool for SaaS teams that scans billing data for churn and MRR leaks, then automates recovery through dunning, cancel-save offers, win-back emails, and onboarding nudges. It is built for founders and bootstrapped teams using Stripe.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
Hype
Hype is a web tool for finding trending YouTube topics by category, time range, and scoring mode. It helps creators spot emerging ideas, inspect source videos, and decide what to cover next.