CLI

The ci-parity commands, their flags, and the exit codes.

Commands

ci-parity check    [--config PATH] [--baseline PATH] [--format text|github|json|sarif]
                   [--strict] [--verbose] [-C DIR]
ci-parity baseline [--config PATH] [--baseline PATH] [--allow-regression] [-C DIR]
ci-parity fix      [--config PATH] [--dry-run] [-C DIR]
ci-parity version

--config and --baseline override the config path and the baseline path from the config. -C changes to DIR before doing anything.
--verbose includes tolerated and allowed findings in text output. JSON always includes them.
version prints ci-parity <version> (<commit>, <date>), with values injected by the build via -ldflags.
fix (from v0.4) rewrites a P001 finding to its suggestion when that cannot change what the step does: its run: value sits on one line (plain, single- or double-quoted, or a literal block with one line); anything before the command is transparent wrappers with their options or NAME=value assignments (from v0.5), with no operator, redirect, quote, expansion, ${{ }} or comment; from the command to the end of the value is exactly the command; a suggestion exists; and the finding is not allowed. Only the command and its arguments are replaced, so wrappers, assignments, quotes, comments and the rest of the file are unchanged (sudo pnpm lint becomes sudo task lint); an edit whose text no longer matches the file, such as an escaped quoted value, is skipped and counted as needing a manual change. It prints <file>:<line>:<col> '<old>' -> '<new>' per edit, then fixed <n> steps in <m> files; <k> inline-tool findings need a manual change, and, when a baseline file exists, a hint to run ci-parity baseline. --dry-run prints would fix and writes nothing. It exits 0, or 2 on a usage or input error.

Exit codes

0: no error-severity findings.
1: at least one error-severity finding, or baseline refused a regression.
2: usage error, config error, unreadable or unparseable YAML, missing Taskfile.

CLI

Commands

Exit codes

See also