CodeLedger FAQs

Question 1

What is CodeLedger and what problem does it solve?

Accepted Answer

CodeLedger is a Claude Code plugin that provides token-level cost intelligence. It solves the problem of opaque AI token usage by tracking costs per-project, per-agent, and per-skill, helping developers understand and optimize their Claude Code expenditures.

Question 2

How does CodeLedger help me optimize my AI coding costs?

Accepted Answer

CodeLedger automatically classifies user vs. overhead token costs, categorizes coding sessions (e.g., generation, debugging), and provides evidence-based cost optimization recommendations with potential dollar savings, guiding you to smarter AI usage.

Question 3

Can I view my usage data in an interactive way?

Accepted Answer

Yes, CodeLedger offers both a local interactive web dashboard with charts, tables, and drill-downs for detailed insights, and conversational querying via 6 MCP Tools, allowing you to ask Claude directly for usage summaries and optimization advice.

Question 4

Is my data private with CodeLedger?

Accepted Answer

Absolutely. CodeLedger is 100% local. All data stays within your `~/.codeledger/` directory in a SQLite database. It performs read-only parsing of Claude Code files, stores only aggregated metrics (no code or prompts), and makes zero network calls, ensuring your privacy and data ownership.

Question 5

What key metrics does CodeLedger track that other tools don't?

Accepted Answer

CodeLedger uniquely provides per-agent token breakdowns, per-skill estimated attribution, automatic user vs. overhead cost classification, and session category classification. This granular detail, combined with optimization recommendations, sets it apart.

CodeLedger

CodeLedger

主な機能

ユースケース

主な機能

ユースケース