ai-eng-audit 0.2.0

Local CLI plotting AI spend against engineering throughput

AI Eng Audit

Plot AI tool spend against engineering throughput on a shared chart. Local, open-source, one command.

What it is

Over the last year a lot of teams burned real money on Claude, Cursor, Copilot, and the rest. The invoices are clear; the output is not.

The tool reads your local git history, PR data, and (optionally) the billing CSV you export from your AI vendor, putting spend and shipped output on the same timeline. You see:

• total AI spend over the window, broken out by month
• PRs merged to the default branch and the L1 ship rate
• PRs opened and then closed, merged and then reverted, or open for too long

How to use it

export GITHUB_TOKEN=ghp_xxxxx
pip install ai-eng-audit
ai-eng-audit scan --repo /path/to/your/repo --window 90d \
    --billing ~/Downloads/anthropic_cost.csv \
    --billing ~/Downloads/openrouter_activity.csv

Python 3.11+. Generate a GITHUB_TOKEN PAT at https://github.com/settings/tokens with repo scope.

--billing can be repeated. Currently supports the Anthropic Console cost export and the OpenRouter Activity export (auto-detected by header). Omit --billing to get the git + PR report only.

Add --format json for JSON output. Metric definitions, supported vendors, and scope-alignment rules are in docs/methodology.en.md.