CLI reference
The ashborn command runs the benchmark and prints the scorecard. It is offline
and deterministic: it makes no network calls and reads only committed fixtures,
so every number reproduces from a clean clone.
Install it, or invoke it directly with npx:
npx @ashborn-sec/cli bench # no install
npm i -g @ashborn-sec/cli # or install the global command
ashborn bench
Commands
ashborn bench
Scores the committed corpus and prints the scorecard — the AUC, recall at a fixed false-positive rate, and per-negative-class figures for each signal. See getting started for how to read the output.
ashborn bench
ashborn bench --json
Emits the same scorecard as JSON, for a CI check or to diff against a stored golden scorecard.
ashborn bench --json
ashborn --version, ashborn -v
Prints the version and exits.
ashborn --version
ashborn -v
ashborn --help, ashborn -h
Shows usage. This is also the output when ashborn is run with no arguments.
ashborn --help
ashborn -h
Exit codes
| Code | Meaning |
|---|---|
0 | Success — bench, --json, --version, or --help ran. |
1 | Unknown command. The help text is written to stderr. |
An unrecognized command exits 1 and prints the usage to standard error; the
recognized commands above all exit 0.
The benchmark makes no network calls and needs no API key. There is no model in the scoring path, so the output is deterministic — the figures the CLI prints are the same figures published in the results, byte-for-byte.