[56steve]

What's new

Breakdown by effort. /token-bar:token breakdown now shows how many requests each model took, and, when the logs record an effort level, the share of usage by effort. Whatever values appear (medium, high, and so on) are shown as they are, so nothing is tied to particular models or levels. The effort section only appears when the logs have it.

cost, optional. /token-bar:token cost shows what your tokens would cost, by model and by effort.

• token-bar ships with no prices: they change, and they differ by account. The first time it prints how to create ~/.claude/token-bar/prices.json with your own prices per million tokens, and lists the model ids found in your logs.
• With prices set it counts input, output, cache writes and cache reads. An id ending in * matches a family of models.
• A model with no price is listed as "no price" and is never guessed. A mistake in the file is explained.
• On a subscription this is not what you pay, since plans are not billed per token. It shows where the expensive usage is, and the output says so.

The scan now keeps the effort level and the four kinds of token for each request, so the scan cache moves to version 2 and is rebuilt once on the first run after updating.

Checks

• 141 plugin tests, 22 Codex tests and the Rust tests pass; ruff is clean.
• Tested on a real week of logs: models with request counts and effort showed correctly, and cost printed the setup message.
• Version bumped to 0.4.0.