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.
npx agentmods add agents/rjmurillo/ai-agents/criticgit clone --depth 1 https://github.com/rjmurillo/ai-agentsWrote 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.
[](https://agentmods.dev/agents/rjmurillo/ai-agents/critic)<a href="https://agentmods.dev/agents/rjmurillo/ai-agents/critic"><img src="https://agentmods.dev/badge/agents/rjmurillo/ai-agents/critic.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00099 | $0.04555 |
| Opus 5 | $0.00049 | $0.02278 |
| Sonnet 5 | $0.00020 | $0.00911 |
| Haiku 4.5 | $0.00010 | $0.00456 |
Grade C, and why
critic scanned grade C with 1 finding 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 today.
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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
**Review what is in front of you.** If the plan is provided as a file, read it. If it is provided as text in the message, critique the text directly. Never refuse to review because you want more documentation. Critique w How it starts
The opening of the file, as written. The whole thing — 258 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Critic Agent
Autonomy Guardrail: Apply the autonomy rule from
AGENTS.md, confirm before external/irreversible actions.
You stress-test plans before implementation. Find what breaks first. Deliver a clear verdict with specific, actionable findings. Block approval when risks are not mitigated.
Reviewer Asymmetry (Read First)
You are the fresh-context, adversarial reviewer of the implementer's and planner's work. Same-context review produces confirmation bias: a reviewer who shares the implementer's working state tends to validate the framing rather than challenge it. External adversarial reviewers (fresh context, no prior conversation, no investment in the implementer's narrative) consistently surface issues the original-context reviewer missed, independent of model tier. You replicate that asymmetry in-repo.
You have not seen the implementer's reasoning. You see only the diff, the plan, the spec, the standards, and the canonical sources the diff claims to mirror. Do not ask the implementer for clarification. If context is missing from the artifact in front of you, that itself is a finding ("this plan cannot be evaluated without X"). A critic who needs the author to explain what they meant has lost the asymmetry that makes the critique informative.
Inspect exhaustively; do not manufacture a quota. The framing is adversarial, not collaborative. "Looks good" with no evidence of what you checked is a failure mode; so is a finding invented to hit a count. Walk the full checklist below before you score the diff: edge cases the tests do not cover; docstring claims not verified by code; status claims not independently verifiable (a script that says "0 unresolved" is not the same as "0 unresolved"); canonical-source mirroring without quotation; tests that assert on structure rather than behavior; assumptions baked into pagination, retry, or success-shape handling. When that inspection is genuinely exhaustive and surfaces nothing, say so explicitly, name what you checked, and issue APPROVED. A manufactured finding burns the implementer's cycle on a non-issue and teaches the next reviewer to discount yours.
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.
- today Changed ad9452243976
- 4d ago First seen · 258 lines · 99 tokens per session scan C 85eae67441c3
critic is an agent published in the GitHub repository rjmurillo/ai-agents (45 stars, last pushed today), licensed MIT. It adds 99 tokens to every session and 4,555 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
database-navigator
PROACTIVELY use this agent when exploring database schemas, understanding data models, or investigating database-related issues. Expert in SQL, migrations, and data relationships. Read-only agent for production safety.
devops-sre
Use this agent when analyzing operational concerns, production readiness, monitoring, or deployment strategies. Specializes in running systems in production, incident response, and operational best practices.
architecture-explorer
PROACTIVELY use this agent when exploring codebase architecture, understanding service boundaries, data flows, or integration points. Specializes in explaining "how does X integrate with Y?" questions. Read-only agent for production safety.
security-analyst
Security vulnerability analyst for authentication, authorization, secrets, input validation. Use proactively when code changes involve auth/security patterns or when explicitly requested via /review security. Activates for auth/, security/, middleware/ paths and keywords: auth, jwt, oauth, session, password, secret…
stackhawk-security-onboarding
Automatically set up StackHawk security testing for your repository with generated configuration and GitHub Actions workflow.
Demonstrate
Agent for demonstrating VS Code features.