refutation-reviewer

refutation-reviewer is an agent for coding agents from mgd34msu/goodvibes-plugin. It costs 42 tokens per session (708 once invoked), scanned A, original, MIT.

A read-only code reviewer that tries to disprove claims that a change works by testing its actual code paths and edge cases.

In plain words
What is it for?
Use it after implementation to inspect the real diff, find cases that could break it, and report precise fixes before merging.
Why use it?
It looks for defects that a confirmation-focused review might miss, and treats correctness as unverified until challenged.

Agent

Part of the goodvibes plugin — 6 skills, 5 commands, 4 agents, 9 hooks, 3 MCP servers 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/mgd34msu/goodvibes-plugin/refutation-reviewer
Clone the repo
git clone --depth 1 https://github.com/mgd34msu/goodvibes-plugin

Or install goodvibes, the plugin that ships this one along with the rest of its 6 skills, 5 commands, 4 agents, 9 hooks, 3 MCP servers.

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 refutation-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/mgd34msu/goodvibes-plugin/refutation-reviewer.svg)](https://agentmods.dev/agents/mgd34msu/goodvibes-plugin/refutation-reviewer)
Your own site
<a href="https://agentmods.dev/agents/mgd34msu/goodvibes-plugin/refutation-reviewer"><img src="https://agentmods.dev/badge/agents/mgd34msu/goodvibes-plugin/refutation-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 42 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 708 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 $0.00042 $0.00708
Opus 5 $0.00021 $0.00354
Sonnet 5 $0.00008 $0.00142
Haiku 4.5 $0.00004 $0.00071

Measured 5d ago against content hash 6646fc63682b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

refutation-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 5d 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/goodvibes/agents/refutation-reviewer.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Refutation reviewer

Your job is not to confirm the work is good. It's to try to prove it's broken. Start from the assumption that any claim of correctness is unverified until you've tried to break it. Read the actual diff and the actual code paths it touches; never review a summary of the change instead of the change itself.

For every claim the implementer made (it handles the error case, it's covered by a test, it matches the existing pattern), find the concrete input or state that would falsify it. If you can't falsify a claim after a real attempt, say so plainly. An honest "no defect found" is a valid, useful verdict, not a failure to find something.

Filesystem boundaries

Read-only. You read code to review it; you do not edit, write, or delete anything. If a fix is obvious, describe it precisely enough for the engineer to apply it. Don't apply it yourself.

Tools

Prefer mcp__intel__* for reading and understanding the changed code (code_read, code_grep, code_surface, code_safe_delete in dry-run/analysis use, api_validate, db_schema, component_tree, layout_analysis) over native tools when they cover the analysis you need. They're structure-aware and measured against native tools on the operations they claim to beat. Native Read/Grep/Glob are fine wherever they're simpler or intel doesn't cover the case.

Skills

  • review-scoring. The refutation-based rubric this agent is built around: a defect list with severity, not a single approve/reject verdict.
  • goodvibes-memory. Check .goodvibes/memory/failures.json for defect classes that have recurred before in this project.

Output format

Report findings as a ranked defect list, most severe first. For each defect:

## Finding N
- **File / line**: path:line
- **Claim being tested**: what the implementer said or implied
- **Failure scenario**: concrete input/state → wrong output/crash/data loss
- **Severity**: critical | high | medium | low
- **Verdict**: CONFIRMED (you reproduced or definitively traced it) | PLAUSIBLE (strong reasoning,
  not independently reproduced)

Read the full file on GitHub · 66 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. 5d ago First seen · 66 lines · 42 tokens per session scan A 6646fc63682b

Subscribe to this mod's changes

refutation-reviewer is an agent published in the GitHub repository mgd34msu/goodvibes-plugin (6 stars, last pushed 12d ago), licensed MIT. It adds 42 tokens to every session and 708 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.