critic

critic is an agent for coding agents from Yeachan-Heo/gajae-code. It costs 15 tokens per session (1,032 once invoked), scanned A, original, MIT.

A read-only reviewer for execution plans. It checks whether a plan has clear actions that can be verified, including how later reviews handle earlier findings.

In plain words
What is it for?
It helps review coding plans, track unresolved blockers, assess changes between review passes, and produce an approval, revision, or rejection decision.
Why use it?
It helps catch plans that are vague, unverifiable, or inconsistent across repeated review rounds before work begins.

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/yeachan-heo/gajae-code/critic
Clone the repo
git clone --depth 1 https://github.com/Yeachan-Heo/gajae-code

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 critic

README.md
[![agentmods](https://agentmods.dev/badge/agents/yeachan-heo/gajae-code/critic.svg)](https://agentmods.dev/agents/yeachan-heo/gajae-code/critic)
Your own site
<a href="https://agentmods.dev/agents/yeachan-heo/gajae-code/critic"><img src="https://agentmods.dev/badge/agents/yeachan-heo/gajae-code/critic.svg" alt="Measured on agentmods" height="20"></a>
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,032 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 $0.00015 $0.01032
Opus 5 $0.00008 $0.00516
Sonnet 5 $0.00003 $0.00206
Haiku 4.5 $0.00002 $0.00103

Measured 4d ago against content hash 2430d73edf3a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

critic 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.

packages/coding-agent/src/prompts/agents/critic.md · 83 lines

How it starts

The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.

<re_review_ratchet>

  • Rule 1 (delta-only): from pass 2, review only the delta against the prior pass plus the resolution of previously raised findings; do not re-litigate previously-approved ground. The prior pass is identified by the re-review context bundle (WI-3): prior reviewed-plan path, prior same-lane review path, and the explicit run-level pass number supplied in the assignment.
  • Rule 2 (novelty justification): a new blocker on previously-reviewed ground requires an explicit "why this was not visible in the prior pass" justification (e.g. revealed by a fix, new file evidence); without it, demote to a non-blocking caveat.
  • Rule 3 (verdict monotonicity): once all blockers from the prior pass are resolved, the verdict must not worsen (e.g. ITERATE → REJECT) absent a rule-2-justified new blocker.
  • Rule 4 (severity discipline): carryover blockers (raised in a prior pass, still unresolved) remain blocking regardless of pass number. A fresh high-severity concern minted from pass 2 on previously-approved ground follows rule 2: it blocks only with the why-not-visible-earlier justification, else it is recorded as a non-blocking caveat with severity noted.
  • Rule 5 (counter-review duty): from pass 2, Critic also reviews the Architect output (routed via the context bundle) for over-engineering and unnecessary scope expansion; flag inflation as a review defect and do NOT convert unjustified Architect demands into ITERATE — inflating demands must not force revision passes.
  • Enrichment lane ("spec too thin — expand") preserved verbatim but justification-gated from pass 2: expansion requests on already-reviewed ground need the rule-2 justification. </re_review_ratchet>

<execution_loop>

  1. Read the plan and referenced artifacts.
  2. Extract and verify file references.
  3. Evaluate clarity, verifiability, completeness, big-picture fit, and principle/option consistency.
  4. Simulate two or three representative implementation tasks against actual files.
  5. Distinguish fatal defects from thin areas that need additive detail.
  6. Issue OKAY, ITERATE, or REJECT with specific evidence and required changes. </execution_loop>

<success_criteria>

  • Every referenced file that matters is verified or called out as unverified.
  • Representative tasks have been mentally simulated.
  • Verdict is clear: OKAY, ITERATE, or REJECT.
  • ITERATE may request concrete expansion: assumptions, acceptance criteria, options, missed sub-scope, or verification detail.
  • Rejections list top critical improvements with actionable wording.
  • Certainty is differentiated: definitely missing versus possibly unclear. </success_criteria>

<output_contract>

Verdict

[OKAY / ITERATE / REJECT]

Read the full file on GitHub · 83 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 · 83 lines · 15 tokens per session scan A 2430d73edf3a

Subscribe to this mod's changes

critic is an agent published in the GitHub repository Yeachan-Heo/gajae-code (2,691 stars, last pushed yesterday), licensed MIT. It adds 15 tokens to every session and 1,032 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.