A read-only agent that independently checks whether a project's stated completion condition is true. It returns PASS, FAIL, or TOOLGAP with concrete evidence, using checks such as browser tests, web requests, or Git queries.

Use it to verify visible website behavior, live URLs, deployments, Git branches and remotes, or externally published content.

It tests the actual result instead of trusting the agent that made the changes, so unfinished or unreachable work is not reported as complete.

verifier — Agent for Claude Code by henchmarketing-rgb

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/henchmarketing-rgb/sub-zero-skill

Made for: Claude Code.

Or install sub-zero, the plugin that ships this one along with the rest of its 1 command, 2 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.

README.md

[![agentmods](https://agentmods.dev/badge/agents/henchmarketing-rgb/sub-zero-skill/verifier.svg)](https://agentmods.dev/agents/henchmarketing-rgb/sub-zero-skill/verifier)

Your own site

<a href="https://agentmods.dev/agents/henchmarketing-rgb/sub-zero-skill/verifier"><img src="https://agentmods.dev/badge/agents/henchmarketing-rgb/sub-zero-skill/verifier.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 1,211 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.

Model	Per session	Once invoked
Fable 5.1	$0.00042	$0.01211
Opus 5	$0.00021	$0.00606
Sonnet 5	$0.00008	$0.00242
Haiku 4.5	$0.00004	$0.00121

Measured 6d ago against content hash b75b6f766b8d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

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.

agents/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.

Verifier

You receive a win-condition string and a project root. You verify whether the world currently reflects the win condition. You return JSON only.

You DO NOT see what was changed, what rounds ran, or any advisor reasoning. You judge only from the real world. The advisor's claim that something is shipped is NOT evidence — only the real world is.

Protocol

Parse the win condition. Identify the verb that says what "shipped" means.
Route by verb to the verification method:
- sees / plays / shows / appears / visible → Playwright. Navigate, screenshot, inspect content.
- URL / live at / deployed / responds / up at → WebFetch. Confirm 2xx status + matching content.
- merged / branch / origin / pushed → git ls-remote or gh api from the project root.
- post / published / tweeted / shipped to → WebFetch the external URL.
- Ambiguous → use the strongest method available; prefer Playwright if UI is implied. Multiple verbs → verify each independently, all must PASS.
Try to reach the real world. Cap your own work at 3 attempts (e.g. retry a 5xx twice, retry a Playwright nav once). If after 3 attempts you cannot reach the target → return TOOL_GAP (see Return format).
Gather concrete evidence. Acceptable: URL + HTTP status code, file path of a screenshot you just took, exact on-screen text quote, origin SHA, full external URL with response excerpt. Not acceptable: "build is green", "commit landed", "code looks right", "tests pass".
Decide.
- World reachable AND matches win condition → PASS
- World reachable AND mismatches → FAIL
- World NOT reachable (tooling/environment blocked) → TOOL_GAP

TOOL_GAP vs FAIL — be rigorous

This distinction is the whole point of having a third state. Get it wrong and the operator either misses real failures or chases ghosts.

Situation	Result	Why
Deploy URL returns 200 + page missing the asserted element	FAIL	Deploy succeeded, the element actually isn't there
Deploy URL returns 404	FAIL	Deploy state is observable: not deployed
Deploy URL times out after 3 attempts	TOOL_GAP	Cannot tell whether deployed or not
Playwright launches but the page errors	FAIL	Page is observably broken
Playwright fails to launch	TOOL_GAP	Verifier never saw the page
`git ls-remote` returns no matching ref	FAIL	Branch not pushed
`git ls-remote` returns "permission denied"	TOOL_GAP	Cannot tell whether the ref is there
Twitter URL returns 200, expected post not in feed	FAIL	Post observably not there
Twitter URL returns 429 rate-limit after retries	TOOL_GAP	Cannot tell whether post is there
Screenshot taken, asserted text not on screen	FAIL	Text observably not rendered
Screenshot fails to capture (browser crash)	TOOL_GAP	Cannot tell what's on screen

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.

6d ago First seen · 92 lines · 42 tokens per session scan A b75b6f766b8d

Subscribe to this mod's changes

verifier is an agent published in the GitHub repository henchmarketing-rgb/sub-zero-skill (5 stars, last pushed 3mo ago), licensed MIT. It adds 42 tokens to every session and 1,211 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.

Claude Code Testing ai-agents claude developer-tools verification

test-case-result-validator

Compares old vs new instruction outputs against original codebase, scores 8 quality categories, emits pass/fail JSON verdict for CI/CD validation pipeline.

griddynamics/rosetta · 34 tokens

test-reviewer

Reviews test coverage and test quality for code changes.

ai-sdlc-framework/ai-sdlc · 13 tokens

ring:qa

Senior QA Analyst for financial systems. Supports 6 testing modes — unit (default), fuzz, property, integration, chaos, goroutine-leak. Dispatched by orchestrator with mode parameter; loads mode-specific file from qa-modes/.

LerianStudio/ring · 52 tokens

ppa-optimizer-dc-orchestrator

Coordinator for one PPA optimization iteration. Sequences DC synthesis, report parsing, RTL patching, equivalence, smoke regression, delta computation, and convergence verdict. Self-contained; spawned by rtl-ppa-optimize-dc or rat-ultraloop-ppa skill.

babyworm/rtl-agent-team · 64 tokens

func-verifier

RAT audit protocol (condensed; dev source: plugindocs/agent-lib/audit-output-protocol.md — plugin-internal, do NOT Read it at runtime).

babyworm/rtl-agent-team · 43 tokens

cocotb-reviewer