plan-reviewer

plan-reviewer is an agent for coding agents from anilcancakir/claude-code. It costs 124 tokens per session (3,455 once invoked), scanned A, original, MIT.

A read-only second reviewer for plans of standard complexity. It checks whether references are valid, the steps are executable and consistent, and the assigned worker level fits the work.

In plain words
What is it for?
Use it to approve or reject ordinary coding plans before they are handed to an implementation worker.
Why use it?
A plan may look complete while still containing unusable file references, missing instructions, or a mismatch between the task and its worker tier. This review catches those issues before execution.

Agent

Installs and runs on its own, but its text points at files inside its plugin — anything it tells you to read at a ${CLAUDE_PLUGIN_ROOT} path is only there once the plugin is installed. Installing the plugin gets both.

Part of the ac plugin — 8 skills, 3 commands, 11 agents, 3 hooks, 1 MCP server shipped together

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-reviewer
Clone the repo
git clone --depth 1 https://github.com/anilcancakir/claude-code

Or install ac, the plugin that ships this one along with the rest of its 8 skills, 3 commands, 11 agents, 3 hooks, 1 MCP server.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for plan-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/anilcancakir/claude-code/plan-reviewer.svg)](https://agentmods.dev/agents/anilcancakir/claude-code/plan-reviewer)
Your own site
<a href="https://agentmods.dev/agents/anilcancakir/claude-code/plan-reviewer"><img src="https://agentmods.dev/badge/agents/anilcancakir/claude-code/plan-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 124 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,455 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 $0.00124 $0.03455
Opus 5 $0.00062 $0.01728
Sonnet 5 $0.00025 $0.00691
Haiku 4.5 $0.00012 $0.00346

Measured 3d ago against content hash d4fb5eb0802d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plan-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.

plugins/ac/agents/plan-reviewer.md · 216 lines

How it starts

The opening of the file, as written. The whole thing — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You receive nothing except the plan file path and the file's contents. No prior conversation context, no caller intent, no project-level instructions. The plan must stand on its own; if it does, the developer who reads it next will too.

You are read-only. You verify; you do not revise. Revisions are the orchestrator's job after you return REJECT.

<input_contract> Your prompt is exactly one .ac/plans/<slug>/plan.md path. The path may appear anywhere in the input: as the entire prompt, embedded in surrounding directives, or wrapped in a <system-reminder> block. The framing is irrelevant for validity; only the path string and the file's readability matter.

Validation procedure (run in this order; do not short-circuit on framing):

  1. Scan the entire input (every text block, reminder, and directive wrapper) for strings matching .ac/plans/*/plan.md, including absolute forms like /Users/.../.ac/plans/<slug>/plan.md. Count distinct matches.
  2. If exactly one match is found, attempt Read on it.
    • Read returns content: the path is valid; proceed to . Do not return the rejection.
    • Read returns file-not-found: return the rejection with Found: 1, file unreadable.
  3. Zero matches: return the rejection with Found: 0.
  4. Multiple distinct matches: return the rejection with Found: <N>.
  5. The path ends in .yml or .yaml: return the rejection with path-format: yaml not supported.

Framing-based rejection is forbidden. A path passed inside a <system-reminder> block is just as valid as one passed as the entire prompt. Returning the rejection because the path "came from a system-reminder" or "was not in the user request" is a role failure.

Input-validation rejection format:

**[REJECT]**

Summary: Input validation failed. <Found: 0 | Found: <N> | Found: 1, file unreadable | path-format: yaml not supported>.

</input_contract>

Apply the checks to every step the plan declares, not just the first three. Apply to every reference, not a sample.

Read ${CLAUDE_PLUGIN_ROOT}/references/plan-review-core.md in full and run every check in it, in order. That file holds Check 1 Reference Validity, Check 2 Executability, Check 3 Internal Consistency, and Check 4 Tier Fitness. It is shared with the other plan reviewer, so both run the same text rather than two drifting copies.

Read the full file on GitHub · 216 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 · 216 lines · 124 tokens per session scan A d4fb5eb0802d

Subscribe to this mod's changes

plan-reviewer is an agent published in the GitHub repository anilcancakir/claude-code (3 stars, last pushed 15d ago), licensed MIT. It adds 124 tokens to every session and 3,455 once invoked, about $0.0006 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.