review(1)

πŸ“¦
version 1.51.0 Β· 2 versions
, in opensuse-leap 16.0

πŸ’… review(1) β€” it's giving code review, bestie

okay so this is the review(1) page from the General Commands Manual, and honestly? it's a whole vibe. it's a local code review tool that runs through goose, and we're gonna walk through it together. grab a snack πŸ“


✨ NAME ✨

review β€” review the current diff using goose. that's it, that's the girl. πŸ‘‘


πŸŽ€ SYNOPSIS πŸŽ€

review [--prompt] [--model] [--provider] [--override-model]
[--turn-limit] [--dry-run] [-q|--quiet] [--no-orchestrate]
[-i|--instructions] [-f|--files] [-c|--check-filter] [-s|--check-scope]
[--checks-only] [--summary-only] [--severity] [-h|--help] [RANGE]

like, yes, that's a lot of brackets. it's fine. we'll get through it together. 🫢


πŸ¦‹ DESCRIPTION πŸ¦‹

so it's local code review. that's the whole concept. πŸ’–

here's the actual tea though β€” it:

  • discovers **/.agents/checks/*.md subagent reviewers (little check besties 🧚)
  • discovers **/.agents/REVIEW.md scoped prompt overrides
  • builds a review request from the working tree (or from an explicit diff range, if you're fancy)
  • and then runs the review through goose

no cap, that's the whole pipeline. ✨


🌸 OPTIONS 🌸

okay this is the long part. buckle up. every single one matters, i promise. πŸ’…

  • --prompt <FILE> β€” path to a Markdown file with a custom base review prompt. this replaces the embedded default prompt, so like... be intentional bestie 😭
  • --model <MODEL> β€” the default model for the main review agent, and also for any check that doesn't declare its own model: in its frontmatter
  • --provider <PROVIDER> β€” the provider for the main review agent
  • --override-model <MODEL> β€” forces every discovered check to use this model, regardless of whatever the check's own model: field says. it's giving "i don't care what you wrote, we're doing it my way" πŸ’€
  • --turn-limit <N> β€” the default turn-limit for orchestrated main-pass subprocesses, and for checks that don't declare their own. heads up though: it does not cap the legacy --no-orchestrate in-process main agent 🫢
  • --dry-run β€” prints the assembled review prompt and the discovered checks instead of actually running the review. perfect for a vibe check before committing to the bit
  • -q, --quiet β€” suppresses non-result output from the underlying agent. quiet queen energy 🀫
  • --no-orchestrate β€” disables the Rust-driven parallel orchestrator and falls back to the single-prompt path, which asks the main agent to delegate each check via delegate(... async: true ...). the default orchestrator dispatches one goose run subprocess per check (capped at 4 concurrent), which bounds wall-clock to the slowest single check instead of waiting on the model to issue dispatches. also: checks with an explicit tool allowlist require the default orchestrator. important. lowkey critical actually πŸ”₯
  • -i, --instructions <TEXT> β€” additional free-form instructions to prepend to the review (think PR intent, commit-message context, "this is a refactor, flag any behavior change"). mirrors amp review --instructions so it's drop-in compatible with existing reviewer wrappers πŸ’–
  • -f, --files <FILE>... β€” restricts the review to a specific set of files. other files in the diff are still passed to the agent for context, but they're excluded from the assembled diff sent to checks. mirrors amp review --files
  • -c, --check-filter <NAME>... β€” only runs checks whose name matches one of these. other discovered checks get skipped. mirrors amp review --check-filter
  • -s, --check-scope <DIR> β€” alternate directory to search for .agents/checks/*.md instead of the repo root. mirrors amp review --check-scope
  • --checks-only β€” skips the main correctness pass and only runs check subagents. mirrors amp review --checks-only
  • --summary-only β€” prints only the diff summary; skips the full review. mirrors amp review --summary-only
  • --severity <LEVEL> [default: medium] β€” minimum severity to display. findings below this rank are dropped from the output. the default is medium, matching Amp's CLI, which hides low from review output. pass --severity low to surface every finding πŸ‘€
  • -h, --help β€” prints help (see a summary with -h)
  • [RANGE] β€” the diff range to review (e.g. "main...HEAD"). defaults to the working tree vs HEAD 🌈

⭐ SEE ALSO ⭐

the whole goose cinematic universe, fr fr β€” go check these out:

ngl, that's a lot of goose. and yet? not one single regret. πŸ’œ

and that's the page!! πŸŽ€ this lil manual entry lives in the General Commands Manual as review(1), and the footer is just review on the left and review(1) on the right β€” cute, symmetrical, iconic. go review some diffs bestie, you're gonna slay πŸ’…βœ¨

πŸ“¦ where this page comes from
  • opensuse-leap 16.0: goose 1.51.0-bp160.1.1 Β· Apache-2.0 AND BSD-2-Clause AND BSD-3-Clause AND CC0-1.0 AND CDLA-Permissive-2.0 AND ISC AND LGPL-3.0-or-later AND MIT AND MIT-0 AND MPL-2.0 AND Unicode-3.0 AND Zlib AND bzip2-1.0.6
  • opensuse-leap 16.0: goose 1.51.0-bp160.1.1 Β· Apache-2.0 AND BSD-2-Clause AND BSD-3-Clause AND CC0-1.0 AND CDLA-Permissive-2.0 AND ISC AND LGPL-3.0-or-later AND MIT AND MIT-0 AND MPL-2.0 AND Unicode-3.0 AND Zlib AND bzip2-1.0.6
  • πŸ’– girlified by deepseek-ai/DeepSeek-V4.1-Flash-1017 on : a language model's rewrite of the original, under its terms