> **Machine-Readable Markdown Mirror** · Canonical HTML: https://senrelay.com/ai-coding-usage-tracker.html
> **Product**: Senrelay (续境 · `qingsongcui/ContextCore`) · **Version**: `v1.0.0-beta.4` · **Updated**: `2026-11-20`
> **Pricing & License**: **$0 (100% Free & Open Source under the MIT License, no signup)** · [Pricing & Limits](https://senrelay.com/pricing.md) · [Extended Spec](https://senrelay.com/llms-full.txt)

# Claude Code and Codex Usage Tracker for Mac — Senrelay

Track observed Claude Code and Codex token usage, retries, and estimated costs on your Mac. Check account quotas separately and export request attempts as CSV.

Updated November 19, 2026 · Senrelay maintainers

## Where do I check tokens, estimated costs, and quotas?

Token usage Review observed input, output, and cache tokens by agent, model, or project. Coverage comes from authorized, parsable native logs and requests seen by Senrelay's local routing. Missing data stays unknown.

Estimated API cost Estimates use the known billing model and rate version and keep currencies separate. Check your provider invoice for actual charges.

Subscription limits and balances Use Account Quotas to refresh supported sources or open the official usage page. Subscription windows and provider balances follow their respective accounts.

## How do I check Claude Code usage and subscription limits?

In Usage & Metrics, select Claude Code and a date range to review the requests and tokens Senrelay has observed. For Claude Code's current subscription windows and reset times, run `/usage` in Claude Code or open its official usage page. [Official Claude Code usage guide](https://code.claude.com/docs/en/costs#using-the-usage-command).

## How do I track Codex usage, account limits, and API cost?

Filter Usage & Metrics to Codex and open Requests for observed usage. Use Read Codex quota in Account Quotas to check the locally signed-in Codex account. For actual OpenAI API charges, check the organization that pays for the API calls. [Official OpenAI API usage dashboard](https://platform.openai.com/usage).

## Why does one request have multiple attempts?

For example, the primary provider can return `429` and a fallback provider can complete the request. The ledger shows **one logical request and two attempts**. Expand it to check the failure, actual model, token counts, and available cost estimates. Export the filtered attempts as CSV for a row-by-row view. [See agent and provider configuration](https://senrelay.com/agent-management.html#provider-switching).

## Frequently asked questions

### Does the tracker cover every Claude Code or Codex call?

Coverage depends on authorized, parsable logs and requests that use the local route. Open Sources and pricing to check the observation sources. Missing usage stays unknown.

### Are estimated costs the charges on my subscription?

Estimated costs use observed tokens and the effective rate. Subscription quotas, actual API charges, and provider balances come from their own sources.

### Which third-party balances or budgets can I check?

Supported sources include authorized DeepSeek and Moonshot Kimi balances and an OpenRouter key budget. OpenRouter account balance needs an authorized management key. Each lookup shows its source and observation time.

## Version and sources

This guide covers Senrelay v1.0.0-beta.4 on Apple Silicon, macOS 14.0 or later. Senrelay is free and open source. Model usage and charges follow your provider's account and billing terms. [Beta.4 release and download](https://github.com/qingsongcui/ContextCore/releases/tag/v1.0.0-beta.4) · [Versioned quota sources](https://github.com/qingsongcui/ContextCore/blob/v1.0.0-beta.4/app/backend/app/services/quota_readers.py) · [Senrelay setup guide](https://senrelay.com/guide.html).
