revet review

The primary analysis command. Scans your code and reports findings.

revet review                    # diff-based: only files changed vs main
revet review --full .           # full repo scan
revet review --fix              # apply auto-fixes
revet review --format json      # machine-readable output
revet review --show-suppressed  # include suppressed findings in output
revet review --ai               # enable LLM reasoning (requires API key)
revet review --gate error:0,warning:5   # fail if gate exceeded
revet review --timings          # print per-analyzer timing breakdown

Flags

Flag	Description
--full	Analyze the entire repository instead of just changed files
--fix	Apply automatic fixes for fixable findings
--format	Output format: terminal (default), json, sarif, github
--fail-on	Exit non-zero if findings of this severity exist: error, warning, info, never
--diff <base>	Diff against this branch/commit (default: main)
--no-baseline	Show all findings, ignoring the saved baseline
--show-suppressed	Show suppressed findings with their suppression reason
--post-comment	Post findings as inline GitHub PR review comments
--module	Run only specific modules (comma-separated, e.g. security,ml)
--ai	Enable LLM reasoning — see AI Reasoning
--max-cost <usd>	Cap AI spend per run in USD (default: $1.00 from config)
--gate <limits>	Override quality gate: error:0,warning:10,info:50 — exit 1 if exceeded
--timings	Print per-analyzer timing breakdown to stderr after the run

Suppressed findings

By default, suppressed findings (inline, per-path, or baselined) are silently filtered out and only counted in the summary. With --show-suppressed, they appear dimmed with a [suppressed: reason] tag — without affecting the exit code or finding counts.

  ⚠️  Possible Hardcoded Password  tests/fixtures/setup.py:8
     [suppressed: per-path rule: **/tests/**]

The summary shows a breakdown by source:

  51 finding(s) suppressed (3 inline, 48 per-path)

AI reasoning

Pass --ai to send each eligible finding to an LLM with a ±4-line code snippet. The model returns a concise note and flags likely false positives. Only warning/error findings without an existing suggestion are sent.

revet review --ai
revet review --ai --max-cost 0.25   # cap spend at $0.25

See AI Reasoning → for setup, model choices, and cost control.

Run log

After each run, the terminal summary shows the command to view the full run log:

  Run log: revet log --show 1772142454966

See revet log for details.