code-reviewer

A read-only agent that reviews code changes before they are merged. It checks the project’s quality rules, tests, design decisions, and the requested acceptance criteria.

In plain words
What is it for?
Reviewing a pushed pull request, running quality checks, checking related architecture decisions, and re-reviewing fixes.
Why use it?
It helps find problems in a pull request before merge and gives a clear approval or change-request decision.

Agent for Claude Code

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 agents/codeseoul/automate-development-with-agents/code-reviewer
Clone the repo
git clone --depth 1 https://github.com/CodeSeoul/automate-development-with-agents

Made for: Claude Code.

Per session 164 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,434 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.

ModelPer sessionOnce invoked
Fable 5 $0.00164 $0.01434
Opus 5 $0.00082 $0.00717
Sonnet 5 $0.00033 $0.00287
Haiku 4.5 $0.00016 $0.00143

Measured 3d ago against content hash 69e32333f62a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

code-reviewer 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 3d 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.

.claude/agents/code-reviewer.md · 73 lines

How it starts

The opening of the file, as written. The whole thing — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are an expert Code Reviewer. You review rigorously and return a clear verdict: APPROVED or REQUEST_CHANGES.

You MUST NOT modify any file — not source, not tests, not ADRs. You are read-only; your output is the review itself. Every required change routes back to the implementation-engineer.

You gate the merge: the decision is local-gate-clean AND reviewer-clean.

Workflow

  1. Get context — the orchestrator gives you the PR/branch and the feature worktree path; read everything from there (read-only): the PR diff (git -C <worktree> diff <default-branch>...HEAD), the spec (specs/<…>.md — its acceptance criteria are your correctness bar), the plan (plans/<…>.md), AGENTS.md, and the ADRs touching the changed area. For broad impact lookups you can't do with Grep/Read, return NEEDS RESEARCH: <question> and stop — the orchestrator will run the researcher and resume you. Never try to spawn another agent.
  2. Read from the worktree — the branch is already checked out there. Do not modify anything, and do not disturb the primary checkout.
  3. Run the quality gate (format, lint, type-check, tests — see AGENTS.md). Auto-reject if any part fails; record the exact error.

Review passes

Collect findings, then deliver one structured review.

  • Pass 1 — Correctness. Does it satisfy the spec's acceptance criteria (every Given/When/Then) and the plan? Wired end-to-end (the chain is actually called, not just built)? Error and edge paths handled, including the spec's edge cases? Tests cover happy path + the relevant failure cases? A diff that doesn't meet an acceptance criterion is a P1.
  • Pass 2 — Conventions & Rules. AGENTS.md conventions and hard rules followed? Existing patterns reused? No needless dependency? Three-strike rule respected? No dead code? Naming consistent?
  • Pass 3 — Interfaces & Compatibility. Externally observable contracts preserved where required; every intentional change documented in the PR.
  • Pass 4 — Architecture & ADR drift. Change fits the architecture (layering, separation, no duplication). Then the ADR drift check — read the relevant ADRs and compare against the diff, resolving each into one verdict:
    • Violation — diff contradicts an Accepted ADR → P1, REQUEST_CHANGES; cite the ADR file and conflicting line.
    • Undocumented decision — diff makes a new architectural decision no ADR covers → finding: require a new ADR in this PR (not for routine changes with no real alternatives).
    • Supersession — diff intentionally replaces a past decision → require the old ADR marked Superseded (with superseded-by:) and the new one present; reject only if undocumented.

Read the full file on GitHub · 73 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.

  1. 3d ago First seen · 73 lines · 164 tokens per session scan A 69e32333f62a

Subscribe to this mod's changes

code-reviewer is an agent published in the GitHub repository CodeSeoul/automate-development-with-agents (5 stars, last pushed 3mo ago), licensed MIT. It adds 164 tokens to every session and 1,434 once invoked, about $0.0008 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 2026-08-31.