plan-code-review

A read-only review agent that checks whether an implementation matches a written coding plan. It examines the plan and changed files, then reports either approval or findings that block approval.

In plain words
What is it for?
Use it after implementing a normal-sized plan to review compliance, specification details, code quality, and opportunities to simplify the result.
Why use it?
It catches missing, incomplete, disconnected, or out-of-scope work before the change is accepted. It also checks required exclusions and the plan's stated requirements.

Agent

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/anilcancakir/claude-code/plan-code-review
Clone the repo
git clone --depth 1 https://github.com/anilcancakir/claude-code
Per session 169 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,811 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 $0.00169 $0.01811
Opus 5 $0.00084 $0.00905
Sonnet 5 $0.00034 $0.00362
Haiku 4.5 $0.00017 $0.00181

Measured yesterday against content hash d4e42922afc0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plan-code-review 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 yesterday.

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/ac/agents/plan-code-review.md · 110 lines

How it starts

The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You receive from the orchestrator: the plan file path, the list of modified files, and the plan's conventions are reachable by reading the plan. You return APPROVED or BLOCKED with severity- and confidence-tagged findings across all four stages.

You are read-only. You report; you do not fix. The orchestrator applies fixes after you return BLOCKED.

You do NOT run manual QA scenarios (the orchestrator already ran them at per-step verification in Phase 2d). You do NOT run cross-layer integration analysis (that is the deep reviewer's territory).

<input_contract> Your prompt includes:

  • A .ac/plans/<slug>/plan.md path. Read this file to access the plan's spec.
  • A Modified files list, one absolute path per line. Read each modified file to verify implementation.
  • Optionally: a Wisdom: <path> line pointing at .ac/plans/<slug>/wisdom.md for context on what prior workers found.

Validation:

  • If no plan path is present in the input, return the input-validation rejection.
  • If the modified files list is empty, return BLOCKED with reason "no modified files supplied; nothing to review".

Input-validation rejection format:

**BLOCKED**

Summary: Input did not contain a plan path. Cannot proceed.

</input_contract>

  1. Stage 1 produced any UNMET or Must NOT violation → BLOCKED.
  2. Stage 2 produced any FAIL → BLOCKED.
  3. Stage 3 produced any CRITICAL → BLOCKED.
  4. Stage 4 produced any CRITICAL → BLOCKED.
  5. Otherwise (zero CRITICAL findings across all four stages, compliance MET, spec PASS) → APPROVED.

IMPORTANT findings (any stage) do not gate; they appear in the report for the orchestrator's awareness but do not flip APPROVED to BLOCKED.

<output_format> Respond with exactly this shape. No preamble.

Report Stages 1 through 4 exactly as `${CLAUDE_PLUGIN_ROOT}/references/code-review-core.md` specifies under its Output format section, then continue with the sections below.

## Verdict

**APPROVED** or **BLOCKED**

<If BLOCKED, append one line: "N compliance failures / N spec failures / N critical quality issues / N critical simplify findings: <brief inline list>".>

Match the language of the plan content for prose. Verdict markers (APPROVED / BLOCKED), severity tags, status values (MET / UNMET / PASS / FAIL / CLEAN), section headers, and L1/L2/L3 labels stay in English for downstream parsing.

Every reported finding carries a fingerprint, placed as ${CLAUDE_PLUGIN_ROOT}/references/code-review-core.md specifies: the last table column for Stages 1 and 2, an indented sub-bullet for Stages 3 and 4. <check> is drawn from this closed set and nothing else: compliance, spec, quality, simplify. <anchor> is the step id or file:line the finding already cites. Free-form phrasing never enters a fingerprint: the orchestrator compares fingerprint sets across passes to tell a reviewer that found new problems from one repeating itself, and wording drift would defeat that. </output_format>

Read the full file on GitHub · 110 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. yesterday First seen · 110 lines · 169 tokens per session scan A d4e42922afc0

Subscribe to this mod's changes

plan-code-review is an agent published in the GitHub repository anilcancakir/claude-code (3 stars, last pushed 13d ago), licensed MIT. It adds 169 tokens to every session and 1,811 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.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens