review-scoring

review-scoring is a skill for Claude Code from mgd34msu/goodvibes-plugin. It costs 66 tokens per session (752 once invoked), scanned A, original, MIT.

A review method that records specific defects and how serious they are instead of reducing a change to one score. It tests each claim in a code change by looking for evidence that could prove it wrong.

In plain words
What is it for?
Reviewing code changes, checking whether tests support their claims, finding defects by file and line, and deciding whether a change is ready to merge.
Why use it?
A pass/fail rating can hide which parts were actually checked. This method makes missing cases, broken behavior, and unsupported claims visible before merging.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the goodvibes plugin — 6 skills, 5 commands, 4 agents, 9 hooks, 3 MCP servers shipped together

Good fit Reviewing code changes, checking whether tests support their claims, finding defects by file and line, and deciding whether a change is ready to merge.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/mgd34msu/goodvibes-plugin/review-scoring
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.

Any agent
npx skills add mgd34msu/goodvibes-plugin --skill review-scoring
Clone the repo
git clone --depth 1 https://github.com/mgd34msu/goodvibes-plugin

Made for: Claude Code.

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 review-scoring

README.md
[![agentmods](https://agentmods.dev/badge/skills/mgd34msu/goodvibes-plugin/review-scoring/github.svg)](https://agentmods.dev/skills/mgd34msu/goodvibes-plugin/review-scoring)
Your own site
<a href="https://agentmods.dev/skills/mgd34msu/goodvibes-plugin/review-scoring"><img src="https://agentmods.dev/badge/skills/mgd34msu/goodvibes-plugin/review-scoring/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for review-scoring

Your own site · 80×15
<a href="https://agentmods.dev/skills/mgd34msu/goodvibes-plugin/review-scoring"><img src="https://agentmods.dev/badge/skills/mgd34msu/goodvibes-plugin/review-scoring.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 66 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 752 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00066 $0.00752
Opus 5 $0.00033 $0.00376
Sonnet 5 $0.00013 $0.00150
Haiku 4.5 $0.00007 $0.00075

Measured 9d ago against content hash f8ea5c4a9e66, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

review-scoring 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 9d 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/skills/review-scoring/SKILL.md · 58 lines

How it starts

The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.

review-scoring

v1's review rubric was a single scalar score (0-10, threshold 9.9) computed by an automated runtime daemon. That daemon is cut for v2 (no standing usage justified keeping a background process for it) and the scalar-score design itself didn't survive review: a single number hides which specific claims were checked and which weren't. v2's rubric is a defect list with severity, produced by a human-in-the-loop review (the refutation-reviewer agent, or you reviewing your own or another agent's work). This skill is that rubric.

The rubric

A review is not "does this look right." It's "what would prove this wrong, and did I check for it." For the change under review:

  1. List the claims. What does the implementer say the change does? (Handles the empty-input case. Covered by a test. Matches the existing error-handling pattern. Doesn't leak the new field to unauthorized users.)

  2. Try to falsify each one. For each claim, find the concrete input, state, or code path that would make it false. Actually trace it or run it, don't just judge plausibility.

  3. Record what survived and what didn't, each with:

    • File/line. Where the defect is, precisely.
    • Failure scenario. The concrete input/state that triggers wrong output, a crash, or data loss. "This looks fragile" is not a failure scenario; "calling this with an empty array throws because arr[0] is accessed unconditionally" is.
    • Severity. critical (data loss/security/crash on common paths), high (wrong behavior on a real path), medium (wrong behavior on an edge case), low (style, maintainability, non-functional).
    • Verdict. CONFIRMED (you reproduced it or traced it definitively) or PLAUSIBLE (strong reasoning, not independently verified). Never present a PLAUSIBLE finding as if it were CONFIRMED.

Grounded checks come first

Before opinion-based review, run what can actually be run: typecheck, the relevant tests, a manual exercise of the changed behavior. A defect a test or typecheck catches outranks a defect found by reading. Cite the tool output, don't restate an opinion a machine already settled.

Read the full file on GitHub · 58 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. 9d ago First seen · 58 lines · 66 tokens per session scan A f8ea5c4a9e66

Subscribe to this mod's changes

review-scoring is a skill published in the GitHub repository mgd34msu/goodvibes-plugin (6 stars, last pushed 16d ago), licensed MIT. It adds 66 tokens to every session and 752 once invoked, about $0.0003 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.