A command that repeatedly reviews code, applies fixes, and verifies the result using a selected large language model. Each pass checks whether issues remain.

Running review passes, fixing findings, and checking builds, tests, or lint results until the review is clean or the pass limit is reached.

It turns review and correction into a repeatable cycle instead of requiring separate manual requests for every round.

review-loop — Command by gopherguides

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/gopherguides/gopher-ai/review-loop

Clone the repo

git clone --depth 1 https://github.com/gopherguides/gopher-ai

Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 3,251 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00016	$0.03251
Opus 5	$0.00008	$0.01625
Sonnet 5	$0.00003	$0.00650
Haiku 4.5	$0.00002	$0.00325

Measured 2d ago against content hash 029befecba1c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

review-loop scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/llm-tools/commands/review-loop.md · 192 lines

How it starts

The opening of the file, as written. The whole thing — 192 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Iterative LLM Review Loop

!if [ ! -x "${CLAUDE_PLUGIN_ROOT}/scripts/setup-loop.sh" ]; then echo "ERROR: Plugin cache stale. Run /gopher-ai-refresh (or refresh-plugins.sh) and restart Claude Code."; exit 1; else "${CLAUDE_PLUGIN_ROOT}/scripts/setup-loop.sh" "review-loop" "REVIEW_CLEAN" 25 "" '{"reviewing":"Resume the review-fix-verify cycle. Run the next review pass.","fixing":"Continue fixing: address remaining review findings, then verify.","verifying":"Continue verification: run build, test, and lint on fixes."}'; fi

1. Parse Arguments

Parse $ARGUMENTS to extract:

--llm <value>: codex (default), gemini, ollama, fable (Claude subagent — no external CLI)
--max-passes <n>: max review passes (default: 5)
--quick: use codex review instead of codex exec (faster, limited to 2-3 findings per pass; codex only)
--tier <value>: gemini service tier (flex/standard/priority; gemini only; default: unset)
Remaining text: scope hint

Store as LLM_CHOICE, MAX_PASSES, QUICK_MODE (default false), GEMINI_TIER, SCOPE_HINT.

Cross-model default: the value of this review is a second model's perspective. When the diff was written by Claude (the usual case), keep the codex default. When the diff was written by Codex (wtcodex flows), prefer --llm fable so a different model family reviews the work.

Persist to .local/state/review-loop.loop.local.json via jq (merge args, pass: 0, quick_mode, gemini_tier) so the stop-hook can restore on re-entry. See ${CLAUDE_PLUGIN_ROOT}/lib/review-loop/state-persist.md for the exact jq invocation; the same pattern repeats in Step 4c.

2. Prerequisite Check

Verify the selected LLM CLI is installed. CRITICAL: Never silently fail or fall back — always present the user with options via AskUserQuestion.

LLM_AVAILABLE=true
case "$LLM_CHOICE" in
  codex)  command -v codex >/dev/null 2>&1 && CODEX_CMD="codex" \
            || LLM_AVAILABLE=false ;;
  gemini) command -v gemini >/dev/null 2>&1 || LLM_AVAILABLE=false ;;
  ollama) command -v ollama >/dev/null 2>&1 || LLM_AVAILABLE=false ;;
  fable)  LLM_AVAILABLE=true ;;  # no CLI — runs as a Claude subagent (see review-phase.md)
esac

Read the full file on GitHub · 192 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 192 lines · 16 tokens per session scan A 029befecba1c

Subscribe to this mod's changes

review-loop is a command published in the GitHub repository gopherguides/gopher-ai (21 stars, last pushed 2d ago), licensed MIT. It adds 16 tokens to every session and 3,251 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Code review

review-loop

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

Iterative LLM Review Loop

1. Parse Arguments

2. Prerequisite Check

What this file has done since we first saw it

Other commands, from other repositories

git

checklist

clarify

specify

analyze

constitution