Claude Code does not document a built-in report that assigns an exact token count to each individual tool call, such as a file read or shell command. To investigate token-heavy activity, use Claude Code’s OpenTelemetry data to find high-token model requests, then correlate those requests with tool events, agents, skills, and large tool results. Treat that correlation as a way to identify likely contributors—not as an exact per-action token ledger.
What you can—and cannot—measure
Claude Code’s OpenTelemetry exports token usage, cost estimates, model-request details, and tool activity. The token counts belong to model requests; tool events describe activity associated with those requests. The official documentation does not promise an exact token bill for each individual tool invocation.
This distinction matters because a tool’s result may become input to a later model request, and a request can include other context as well. Anthropic’s tool-use pricing explanation describes how requests can account for input and output tokens, tool definitions, tool-use blocks, and tool-result content. A large tool result can be a useful lead, but it does not establish how many billed tokens should be assigned to that tool alone.
Set up OpenTelemetry for Claude Code
Use the official Claude Code monitoring guide to enable telemetry and configure exports. Its quick start uses CLAUDE_CODE_ENABLE_TELEMETRY=1 and metrics and logs exporters such as OTLP. You also need a configured collector or monitoring backend able to receive the exports.
#1 Best Overall
- AI CODING USAGE AT A GLANCE Dedicated WiFi desk display for Claude Code users showing your 5 hour session usage and weekly usage as clean color gauges.
- RUNS ON ITS OWN OVER WIFI No laptop app, no browser tab, and no background daemon needed. Once connected, codeMeter updates automatically from your own Claude account.
- FAST GUIDED SETUP Ships pre flashed and ready to configure. Plug in, join the setup WiFi, enter your home WiFi, paste your Claude Code token, and start tracking.
- COLOR LIMIT WARNINGS Easy visual cues shift from green to amber to red as you approach usage limits, with reset countdowns to help you plan your work.
- MULTIPLE VARIATION OPTIONS Choose from available finish and power options, including USB powered and battery equipped versions depending on the selected model.
For an individual investigation, filter the telemetry to the relevant user, session, or prompt where those attributes are available. For organization-wide analysis, the same data can help compare usage across users and teams. What you can retain and inspect depends on your collector and backend configuration.
Find the requests with the highest token use
Start with the claude_code.token.usage metric. The monitoring guide documents dimensions that can help narrow usage by token type, user, team, model, skill.name, plugin.name, or agent.name. Use them to locate concentrations—for example, whether high usage clusters around a particular agent or model—before investigating individual request activity.
Rank #2
- AI CODING USAGE AT A GLANCE Dedicated WiFi desk display for Claude Code users showing your 5 hour session usage and weekly usage as clean color gauges.
- RUNS ON ITS OWN OVER WIFI No laptop app, no browser tab, and no background daemon needed. Once connected, codeMeter updates automatically from your own Claude account.
- FAST GUIDED SETUP Ships pre flashed and ready to configure. Plug in, join the setup WiFi, enter your home WiFi, paste your Claude Code token, and start tracking.
- COLOR LIMIT WARNINGS Easy visual cues shift from green to amber to red as you approach usage limits, with reset countdowns to help you plan your work.
- MULTIPLE VARIATION OPTIONS Choose from available finish and power options, including USB powered and battery equipped versions depending on the selected model.
Then inspect claude_code.llm_request spans and api_request events. These expose request-level counts for input, output, cache-read, and cache-creation tokens. Where present, use identifiers such as prompt.id, session.id, and request_id to connect requests with the relevant prompt or session.
Include cached input in the total
The metric’s input type excludes cached input. For a fuller total input calculation, add the request’s input_tokens, cache_read_tokens, and cache_creation_tokens. Ranking requests using only uncached input_tokens can make cache-heavy activity appear smaller than it is.
Rank #3
- AI CODING USAGE AT A GLANCE Dedicated WiFi desk display for Claude Code users showing your 5 hour session usage and weekly usage as clean color gauges.
- RUNS ON ITS OWN OVER WIFI No laptop app, no browser tab, and no background daemon needed. Once connected, codeMeter updates automatically from your own Claude account.
- FAST GUIDED SETUP Ships pre flashed and ready to configure. Plug in, join the setup WiFi, enter your home WiFi, paste your Claude Code token, and start tracking.
- COLOR LIMIT WARNINGS Easy visual cues shift from green to amber to red as you approach usage limits, with reset countdowns to help you plan your work.
- MULTIPLE VARIATION OPTIONS Choose from available finish and power options, including USB powered and battery equipped versions depending on the selected model.
Connect high-token requests to tools and agents
Once you have identified a high-token request, inspect nearby or linked tool events. Claude Code telemetry includes tool names, execution duration, and result-size fields. The tool_use_id links a tool event to related tool records; use it, along with available request or session identifiers, to examine which tools ran around the request.
Look for recurring tool activity, unusually large results, failures, or errors. These details can help explain why a request may have needed substantial context or follow-up. They are clues, not proof that the tool caused all the request’s tokens: a model request can contain other context, and telemetry describes tool-result size in bytes and approximate result-token size rather than a definitive per-tool billed count.
Rank #4
- Easy Tracking: Easily track your initiative and quickly change your initiative’s order by removing and sliding your markers into place, ensuring smooth and organized gameplay.
- Marker Compatibility: The white acrylic markers can be written on with dry erase markers, allowing for easy and quick changes to the initiative order.
- Flexible Setup: Measuring 11” x 5.25” x 0.75”, this tracker can be used both standing vertically and laying flat, providing flexibility in how you arrange your game space.
- Versatile Usage: Includes three different colored pegs to track rounds and events, enhancing gameplay clarity and organization.
- Made in the USA: Crafted from top-quality durable Birch wood, these stands are designed and manufactured in the USA, ensuring superior craftsmanship.
Choose the right source for the question
| Question | Use | What it establishes |
|---|---|---|
| Which requests or users have the most token usage? | Claude Code OpenTelemetry metrics and request spans/events | Token counts by request and documented dimensions; the available detail depends on the exported data. |
| Which tools may be contributing to a heavy request? | Correlated tool events and their result-size and execution fields | Tool activity and useful investigative clues, not an exact token allocation for each tool call. |
| What amount was officially billed? | The relevant API provider’s billing records | Authoritative billed amounts. Claude Code’s cost telemetry is an approximation. |
Why CLI output is not a per-action token leaderboard
The Claude Code CLI reference documents print mode (-p), JSON and stream-JSON output formats, verbose turn-by-turn output, and the print-mode --max-budget-usd limit. It does not document a built-in leaderboard assigning exact token counts to individual tool calls. Check the installed version’s current help output before relying on commands mentioned in older guides.
JSON output, stream-JSON output, and the Agent SDK can support scripted integrations. The SDK overview describes programmatic access to the agent loop and Claude Code capabilities, but that description does not establish a command for ranking per-tool token use.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
