You write code. You lose patience. You say please.
RageReport turns your coding-agent history into
receipts for all of it.
Free. Open source. Stays on your machine.
RageReport — your keyboard has stories
Usage: ragereport [scan|cost|slop|report|doctor] [options]
Commands:
scan Count your swears, mild insults, and polite expressions (default)
cost Estimate token costs and write an interactive HTML cost report
slop Count common writing patterns in assistant responses
report Write a combined standalone HTML report
doctor Check history locations, formats, and read errorsAnd nine local agent histories.
You already know what you built.
Here’s how it went down.
From “please fix this” to the third stretched-out fuuuuck. Count swears and polite expressions, find your favorite words, and compare your projects.
English + phonetic Hindi$ npx ragereport scan --weekFollow input, output, reasoning, and cache tokens across agents and models. Missing usage stays missing. An estimate stays an estimate.
Token usage + cost breakdowns$ npx ragereport cost --month --offlineYour AI has habits too. Spot stock phrases, reflexive agreement, and formatting tics. A writing-style heuristic, with a sense of humor.
Patterns, not a quality score$ npx ragereport slop --weekA quick terminal scan or the whole browser report.
Use the same local history. Ask a different question.
npx ragereport scan --weekSwears, please-and-thank-yous, spelling variants, and your rage ratio. English and phonetic Hindi included.
Your conversations stay on your computer. RageReport reads your history without changing it, and language scans work entirely offline. Use --offline for cost reports too.
No. RageReport reads local history files without changing them. There is no account, API key, or transcript upload. Language scans work offline. Cost reports can refresh a public price catalog; use --offline to keep those offline too.
Claude Code, Codex, Cursor, OpenCode, Amp, Cline, Pi, T3 Code, and Zed. Available messages and usage depend on each agent’s local history format. The doctor command helps you find missing or unsupported histories.
It matches Latin phonetic spellings such as chutiya, bhosdike, and shukriya, alongside English. It groups spelling variants and stretched words, skips code and hostnames, and keeps mild insults separate from swear counts.
Usually they are API-equivalent estimates based on available usage and model prices. Recorded charges are shown separately where an agent provides them. Subscription charges, missing usage, and unknown prices cannot be reconstructed from a local history.
HTML reports and SVG summary cards contain aggregate counts and word variants, without transcripts, message excerpts, or full local paths. Review them before sharing. CLI JSON and doctor diagnostics may include local paths.
Run npx ragereport with Node.js 24.14 or newer. No global install or pnpm is needed to use the published CLI. Start with npx ragereport --help for all commands and options.
One command. Your own numbers. Your history stays local.
Your sense of humor is optional.