Terminal report

Layout

The terminal report starts with the tool version and groups findings by rule, ordered by first-participant file within each rule. Blank lines separate rule headings and findings.

• Each finding shows its severity and message, followed by a --> primary location and related: locations. Every location retains its full test title and path:line:column.
• Baselined findings remain visible with a [baselined] marker and their original severity.
• Incomplete-run and changed-file warnings appear before findings.

Each finding ends with a rule-specific action: line:

• Name collisions call for renaming.
• Parameterization preserves every case and assertion.
• Similarity and subsumption require review before changing tests.
• Exact duplicates prompt review of removal or repair of intended coverage; cross-file duplication may be deliberate.

Actions also remain visible for baselined findings. A rule configured at severity off is not computed, so a CLI run never reports a finding for it.

Footer

The footer summarizes analyzed tests and files, displayed severities, inline suppressions, and baseline debt. Truncated reports label shown findings and omitted counts explicitly. The severity summary keeps its fixed errors, warnings, off shape, so the off count of a CLI run is always 0.

The CLI also shows the existing error-level duplication threshold: unique unbaselined error finding locations divided by analyzed executable tests. Locations may belong to suites or hooks too, so the numerator is labeled as locations rather than tests. Zero executable tests have no rate to display, even when a suite finding fails policy.

Analysis policy is separate from the threshold and includes --fail-on-new and the selected incomplete-run policy. PASS can be qualified by incomplete coverage; it does not certify the whole repository. This is the analysis decision before report writing: a later output error still exits operationally.

Library write_report omits threshold and policy labels; callers can supply TerminalReportContext to write_report_with_context.

Detection time

The CLI appends Detection time: N.N ms after the finding count, matching the sum of monotonic discovery, primary parsing/extraction and analysis durations.

• Discovery includes project resolution and globals detection.
• Baseline preparation (including a ref-baseline analysis), baseline writes and report generation/I/O are excluded.
• --no-metrics gives byte-stable terminal output across equivalent runs. The Action exposes the same no-metrics input (default false). It is a CLI presentation option, not a configuration-file key.
• Machine reports contain no timing fields and are unchanged by the flag.
• Library reports omit timing unless the caller supplies DetectionTiming through TerminalReportContext::with_timing.

Color

Color is enabled only when stdout is a terminal. Set NO_COLOR to any nonempty value to disable it; an empty value allows color.

• Rule headings are bold cyan, error and warning labels red and yellow, locations blue, and summary counts gray.
• A library caller can render a finding at severity off; its label is gray. The CLI never produces one.
• Piped output and machine or file reports contain no tool-added ANSI styles.
• Library reporters default to plain text; TerminalReporter { no_color: false } explicitly enables styles.
• On Windows, the CLI enables virtual-terminal processing on stdout before adding styles; if console access or enabling fails, it uses plain text instead. Opting out of color and machine-only reporting do not change console modes.

Back to Reports.