skills.rest
techwolf-ai
techwolf-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

token-doctor

Analyze local Claude usage transcripts to identify spending patterns and cost drivers.

95|2|Updated Feb 27, 2026
Search Tags:#codex#diagnosis#claude-code#transcripts#cost-analysis#token-cost#markdown-report
One-click install
npx skills add https://github.com/techwolf-ai/ai-first-toolkit --skill token-doctor
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: token-doctor
Source: https://github.com/techwolf-ai/ai-first-toolkit/tree/main/plugins/ai-adoption/skills/token-doctor
Command: npx skills add https://github.com/techwolf-ai/ai-first-toolkit --skill token-doctor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Claude Code and Codex transcripts can hide cost drivers and session inefficiencies. token-doctor provides a local, two-stage diagnosis of spending by analyzing transcripts, projecting per-project spend, and offering a deeper Haiku-based investigation with parallel subagents to reveal actionable improvement opportunities.

Core Features & Use Cases

  • Two-stage diagnosis: quick terminal report plus optional hotspot deep dives.
  • Per-project spend insights: identify which cwd/project carries the most cost, marathon vs zombie vs cache rebuilds.
  • Generates a Markdown report for sharing, auditing, and archival.

Quick Start

Run the token-doctor on your local Claude transcripts to generate a cost diagnosis.

Frequently Asked Questions about token-doctor

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I analyze Claude Code token costs from local transcripts?▼

Analyze Claude Code token costs by running a local two-stage diagnosis on your transcripts to identify spending patterns, map usage to token prices, and generate a quick terminal report.

What is the best way to find cost drivers in Claude Code and Codex sessions?▼

Find cost drivers by applying a Haiku-based deep dive with parallel subagents that fan out across session hotspots to reveal marathon, zombie, and cache rebuild inefficiencies.

Can I generate a markdown report for Claude Code token usage auditing?▼

Yes, you can generate a markdown report for token usage auditing that details per-session and per-project diagnostics for sharing and archival.

Does token-doctor process Claude transcript data locally without sending it to a remote server?▼

Yes, the diagnosis satisfies safety constraints by keeping all transcript data processing local while mapping usage to token costs via the repository pricing module.

How do I identify which project directory has the highest Claude token spend?▼

Identify high-cost project directories by analyzing local transcripts to map per-project spend insights, revealing which working directory carries the most token cost.

Related Skills

Explore More Software Engineering Skills→
wangqianCAIwangqianCAI

model-usage

Summarize CodexBar JSONL cost logs for Codex and Claude providers.

Community
Intermediate
aalmadaaalmada
15

scaffold-composite-projection

Build a Composite Projection in Marten to chain multiple projections.

Community
Advanced
thewoolleymanthewoolleyman

V3 Deep Integration

Integrate claude-flow into agentic-flow@alpha with an adapter layer and phased migration.

Community
Advanced
Cure-Consulting-GroupCure-Consulting-Group

finops

Manage FinOps processes for Firebase and GCP projects with budgeting and cost allocation.

Official
Advanced