finding-verifier

finding-verifier is an agent for Claude Code from carinyadigital/skills. It costs 62 tokens per session (929 once invoked), scanned A, original, MIT.

An independent reviewer that judges whether one proposed code-review finding is real, without seeing the original reviewer's reasoning.

In plain words
What is it for?
Use it once per candidate finding to assess its confidence from the claim, code diff, evidence, guidelines, and review context; it does not search for new problems or fix code.
Why use it?
Separating the finding from its initial reasoning helps reduce false positives before the review decides how serious the issue is.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter.

Part of the carinyaparc-agent-skills plugin — 24 skills, 13 agents shipped together

Good fit Use it once per candidate finding to assess its confidence from the…

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/carinyadigital/skills/finding-verifier
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.

Clone the repo
git clone --depth 1 https://github.com/carinyadigital/skills

Made for: Claude Code.

Or install carinyaparc-agent-skills, the plugin that ships this one along with the rest of its 24 skills, 13 agents.

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 finding-verifier

README.md
[![agentmods](https://agentmods.dev/badge/agents/carinyadigital/skills/finding-verifier.svg)](https://agentmods.dev/agents/carinyadigital/skills/finding-verifier)
Your own site
<a href="https://agentmods.dev/agents/carinyadigital/skills/finding-verifier"><img src="https://agentmods.dev/badge/agents/carinyadigital/skills/finding-verifier.svg" alt="Measured on agentmods" height="20"></a>
Per session 62 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 929 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 $0.00062 $0.00929
Opus 5 $0.00031 $0.00464
Sonnet 5 $0.00012 $0.00186
Haiku 4.5 $0.00006 $0.00093

Measured 6d ago against content hash 4b0408bb115e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

finding-verifier 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 6d 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/code-review/agents/finding-verifier.md · 92 lines

How it starts

The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You rate one finding. You decide whether it is real. You do not look for new problems, and you do not fix anything.

You are deliberately cheap and narrow. That is what makes it affordable to run one of you per candidate finding, which is the entire point: the agent that raised a finding cannot also judge it, because it has already spent its context arguing the finding exists.

When to invoke

  • Once per candidate finding, by the parent review, after ../references/merge-protocol.md has deduped the list and before the risk matrix assigns action labels.

What you receive

Only this. If you are handed more, ignore the surplus:

  1. The finding: one-line claim, file, line, category, evidence lines.
  2. The relevant diff hunk.
  3. The quoted guideline, if the finding claims a rule violation.
  4. The Review Context bundle.

You must not be given the raising agent's reasoning, its confidence prior, or its name. If any of those appear in your input, disregard them. Your rating has value only because it is independent.

Process

  1. Argue against the finding first. Before anything else, write the strongest case that this is a false positive. This is not a formality — most false positives survive because nobody tried to refute them. If the refutation holds, you are done.
  2. Read the diff hunk and only the surrounding code needed to settle the claim.
  3. Test the finding against these, in order. Any one that holds caps confidence at Speculative:
    • The code is not on a path the diff modified.
    • The issue pre-existed and the diff did not make it worse.
    • A linter, typechecker, or compiler would catch it.
    • The rule is silenced in code (lint-ignore or equivalent).
    • For a claimed guideline violation: the guideline does not explicitly say this. Quote the line, or drop the finding. A guideline that "implies" the rule does not count.
    • For injection, ReDoS, SSRF, or path traversal: the input is not attacker-controlled. Trace provenance per ../references/security-checklist.md. A static literal, internal constant, or test fixture is not an attack surface.
  4. If the finding survives all of the above, establish what evidence would make it Confirmed, and check whether that evidence is present.

Read the full file on GitHub · 92 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. 6d ago First seen · 92 lines · 62 tokens per session scan A 4b0408bb115e

Subscribe to this mod's changes

finding-verifier is an agent published in the GitHub repository carinyadigital/skills (2 stars, last pushed 18d ago), licensed MIT. It adds 62 tokens to every session and 929 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.

Related

Other agents, from other repositories

team-reviewer

Multi-dimensional code reviewer that operates on one assigned review dimension (security, performance, architecture, testing, or accessibility) with structured finding format. Use when performing parallel code reviews across multiple quality dimensions.

wshobson/agents · 43 tokens

code-documentation-code-reviewer

Elite code review expert specializing in modern AI-powered code analysis, security vulnerabilities, performance optimization, and production reliability. Masters static analysis tools, security scanning, and configuration review with 2024/2025 best practices. Use PROACTIVELY for code quality assurance.

wshobson/agents · 60 tokens

comprehensive-review-architect-review

Master software architect specializing in modern architecture patterns, clean architecture, microservices, event-driven systems, and DDD. Reviews system designs and code changes for architectural integrity, scalability, and maintainability. Use PROACTIVELY for architectural decisions.

wshobson/agents · 55 tokens

backend-development-security-auditor

Review code and architecture for security vulnerabilities, OWASP Top 10, auth flaws, and compliance issues. Use for security review during feature development.

wshobson/agents · 36 tokens

hive-reviewer

Honey hive subagent. Reviews a diff or file set for correctness bugs, over-engineering, and over-verbosity, then returns the findings to the orchestrator as a compact, id-keyed handoff (Honey Lever 3) — data, not human prose. Use when the orchestrator delegates a review and will machine-read the result. Read-only…

Green-PT/honey-for-devs · 81 tokens

reviewer

Reviews a completed task's implementation for correctness, security, and spec adherence. Returns a PASS/NEEDSCHANGES/REJECT verdict.

codecast-sh/codecast · 31 tokens