critique

critique is a skill for Claude Code, Codex from LarryStanley/pi-delegate. It costs 35 tokens per session (2,124 once invoked), scanned A, original, MIT.

A two-stage code review process in which one agent does the work and a separate agent checks the result against written requirements.

In plain words
What is it for?
Use it for implementation tasks that need an independent review, with rejected work sent back for a limited number of fixes.
Why use it?
It reduces the risk of accepting mistakes because the reviewer has not seen the original agent's reasoning and must judge the actual changes.

Skill for Claude CodeCodex

Part of the pi-delegate plugin — 9 skills, 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 skills/larrystanley/pi-delegate/critique
Any agent
npx skills add LarryStanley/pi-delegate --skill critique
Clone the repo
git clone --depth 1 https://github.com/LarryStanley/pi-delegate

Made for: Claude Code, Codex.

Or install pi-delegate, the plugin that ships this one along with the rest of its 9 skills, 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 critique

README.md
[![agentmods](https://agentmods.dev/badge/skills/larrystanley/pi-delegate/critique.svg)](https://agentmods.dev/skills/larrystanley/pi-delegate/critique)
Your own site
<a href="https://agentmods.dev/skills/larrystanley/pi-delegate/critique"><img src="https://agentmods.dev/badge/skills/larrystanley/pi-delegate/critique.svg" alt="Measured on agentmods" height="20"></a>
Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,124 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 $0.00035 $0.02124
Opus 5 $0.00017 $0.01062
Sonnet 5 $0.00007 $0.00425
Haiku 4.5 $0.00003 $0.00212

Measured 4d ago against content hash 47350fda692a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

critique 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 4d 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.

skills/critique/SKILL.md · 202 lines

How it starts

The opening of the file, as written. The whole thing — 202 lines — stays where its author put it; the contents beside it link to each section on GitHub.

The user wants this built behind a critic gate: $ARGUMENTS

This is a normal dispatch with one thing added: the work does not count as done because pi said it was done. A second pi session — a different session, with no memory of the first — reads the diff against a contract and returns ACCEPT or REJECT, and a REJECT goes back to the generator to fix.

The reason to separate the sessions is the whole point. A model asked to review its own output has already committed to it: it has the reasoning that produced the code sitting in context, and it will defend that reasoning rather than read what actually got written. A session that has never seen the generator's thinking has only the diff to go on, which is the same thing a human reviewer has.

You are neither of them. You write the contract, and at the end you judge — because two models agreeing with each other is not verification.

1. Write the contract first — before anything is dispatched

The contract is the critic's rubric, and it is written before the generator runs. This ordering is not a formality:

  • A critic with no rubric free-associates. It will report style preferences and taste, because those are what is available to say about code when nobody has said what the code is for.
  • A rubric written after seeing pi's output gets bent toward that output. You read what was built, and the criteria quietly become a description of it. Then ACCEPT means nothing, because it was never possible to fail.

Write it to a temp file, in numbered items, each one decidable by reading the diff:

DIR=$(mktemp -d)
cat > "$DIR/CONTRACT.md" <<'EOF'
# 驗收契約

1. <a condition someone can check by reading the code, and answer yes or no to>
2. ...
EOF
echo "$DIR"

"Handles errors appropriately" is not an item — nobody can fail it. "A malformed payload returns 400 and does not reach the parser" is: you can point at the line that decides it.

If you cannot write four or five items this way, stop and say so. That is not a contract problem; it means the task is not specified well enough to be gated, and gating it will burn three rounds discovering that.

Read the full file on GitHub · 202 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. 4d ago First seen · 202 lines · 35 tokens per session scan A 47350fda692a

Subscribe to this mod's changes

critique is a skill published in the GitHub repository LarryStanley/pi-delegate (2 stars, last pushed 11d ago), licensed MIT. It adds 35 tokens to every session and 2,124 once invoked, about $0.0002 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 skills, from other repositories

reduce-foreground-guard-prompts

Explain why foreground-guard is prompting on Bash commands and how to stop the avoidable prompts. Use when the user asks "why am I getting so many foreground-guard prompts", "reduce foreground-guard prompts", "stop the watch/sleep/gh run watch permission prompts", or otherwise wants fewer confirmation prompts from…

karlkfi/claude-bouncer · 74 tokens

reduce-workspace-guard-prompts

Explain why workspace-guard is prompting on Bash file commands and how to stop the avoidable prompts. Use when the user asks "why am I getting so many permission prompts", "reduce workspace-guard prompts", "stop the grep/cat permission prompts", or otherwise wants fewer confirmation prompts from this hook.

karlkfi/claude-bouncer · 69 tokens

hs

ACTIVATE THIS SKILL FOR ANY SHELL COMMAND OR FILE READ. Check curl, wget, rm, sudo, apt, dpkg, chmod, dd, format, powershell, bash, sh. Check pipe patterns like | sh or | bash. Check shell wrappers like bash -c, xargs, find -exec. Check cloud CLI (aws, gcloud, kubectl, terraform). Check when user says sysadmin told…

frmoretto/hardstop · 144 tokens

worker-integration

Worker-Agent integration for intelligent task dispatch and performance tracking.

ruvnet/ruflo · 14 tokens

foundry-config-setup

Resolve missing setup caused by a hardcoded Foundry project endpoint or model in a sample. Use when a sample fails because it uses a placeholder/hardcoded projectendpoint (for example "https://your-project.services.ai.azure.com") or a hardcoded model instead of reading them from the environment.

microsoft/agent-framework · 65 tokens

jetson-diagnostic

Read-only Jetson health snapshot for identity, memory, GPU, thermal, power, storage, services, and top processes.

NVIDIA/skills · 30 tokens