Commands

Run cargo judge --help or cargo judge <command> --help for the complete option list of the installed version.

Combined report

Command	What it does
cargo judge	Combined findings from every detector that needs no opt-in configuration, grouped for triage. --details lists every finding. --progress PATH writes JSON Lines phase events.

Workspace context

Command	What it does
inspect	Crates, source files, entry points and available tiers, via cargo metadata.
structure	Crates, targets, internal dependency edges and measured structural concentration.
complexity	Production and test complexity rankings from the checked-out tree.
map	Compact facts for refactoring plans: files by aggregate cyclomatic complexity and the largest clone families. --include-tests adds test-only code.
impact PATH	Crate targets and judge commands that read one Rust source file directly. It does not predict findings.

Focused analyses

Command	What it does
health [--score]	Complexity, syntax-level slop signals and an optional 0–100 health score with letter grade.
dupes --mode strict|mild|weak|semantic	Duplicated token spans grouped into clone families. Default mild, minimum 20 tokens; test code with --include-tests.
deps	Dependency-kind hygiene and local name-collision checks. Opt-in: --check-crates-io, --check-rustc-lints, --audit-json PATH. --why CRATE explains one dependency instead.
api	Focused public API surface analysis (api-surface remains available).
boundaries	Crate boundaries from judge.toml plus dependency cycles. --graph dot|mermaid prints the crate graph instead.
module-graph	unlinked-file and orphan-module findings from each crate’s real mod tree.
errors, tests, unsafe, slop	Focused current-state views with concrete locations and stated Fast-Tier limits.
coverage	Imports an existing cargo-llvm-cov LCOV report and flags untested-hotspot functions. judge never measures coverage itself.

Refactoring support

Command	What it does
refactor [PATH]	Deterministic, evidence-backed review queue. With --format json it adds a ranked next_actions summary (top 10). It never writes a patch.
patterns, principles	Heuristic design-pattern and design-principle readings. Advisory only; they never affect the verdict or exit code.
explain-pattern, explain-principle	Full evidence, preconditions and contraindications of one candidate.
fix-preview	The migration plan and affected call sites of a pattern candidate — deliberately no patch.
explain-rule ID	Documentation of one rule from the static registry. See Rules.

Baselines

Command	What it does
--save-baseline	Saves the current findings to .judge/baseline.json.
--baseline PATH	Compares a report with a saved baseline.
compare BASELINE	Compares the current project state with a saved baseline artifact. Never reads Git history.

Deep Tier (--features deep)

Command	What it does
dead-code [--include-tests]	pub items with no reference from another workspace crate and no reachability from an entry point of their own crate.
explain <item-path> --why-live	The shortest evidenced call path from an entry point to the item.

Global options

Option	Effect
--format tty|json|sarif|markdown	Output format. See Output formats.
--output PATH	Artifact path instead of the default .judge/<command>.json or .judge/judge.md.
--color auto|always|never	Colour policy of the combined report. auto honours NO_COLOR.
--details	Every finding in the combined terminal report.

Exit codes

Code	Meaning
0	Clean. Commands without a verdict always exit 0. A closed stdout (broken pipe) also exits 0.
1	A baseline verdict failed on introduced gating findings.
2	A real error: broken judge.toml or baseline, analyzer or toolchain failure, an unavailable health score, or incomplete analysis.