What is adversarial-qc?

A structured review process for checking AI-produced reports, plans, analyses, scripts, and other substantial work. It uses one or two independent reviewers and compares their findings.

Use it to review a deliverable against a standard or custom checklist and produce a quality-control certificate or written summary.

It helps catch unsupported claims, omissions, and disagreements before a deliverable is sent to someone else.

adversarial-qc — Skill for Claude Code and Codex by LegalQuants

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 LegalQuants/lq-skills --skill adversarial-qc

Clone the repo

git clone --depth 1 https://github.com/LegalQuants/lq-skills

Made for: Claude Code, Codex.

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/skills/legalquants/lq-skills/adversarial-qc/github.svg)](https://agentmods.dev/skills/legalquants/lq-skills/adversarial-qc)

Your own site

<a href="https://agentmods.dev/skills/legalquants/lq-skills/adversarial-qc"><img src="https://agentmods.dev/badge/skills/legalquants/lq-skills/adversarial-qc/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/legalquants/lq-skills/adversarial-qc"><img src="https://agentmods.dev/badge/skills/legalquants/lq-skills/adversarial-qc.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 138 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 2,853 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	$0.00138	$0.02853
Opus 5	$0.00069	$0.01426
Sonnet 5	$0.00028	$0.00571
Haiku 4.5	$0.00014	$0.00285

Measured 12d ago against content hash 4f94fb901dc6, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

adversarial-qc 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 12d 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/adversarial-qc/SKILL.md · 178 lines

How it starts

The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Adversarial QC

Structured quality control for AI deliverables. Two agents independently verify a deliverable against a checklist, then results are compared. Agreements = high confidence. Disagreements = flagged for human review.

Quick Start

Read the deliverable to be reviewed
Read references/checklist.md for the verification checklist
Read references/config.md for model configuration
Run the QC process (see Workflow below)
Generate the QC certificate

Configuration

The user controls these settings. Ask if not specified:

Mode: single (one model, two agent personas) or cross (two different models)
Models: Which model(s) to use. Defaults: primary = current session model, secondary = user's choice
Checklist: standard (general-purpose) or a custom checklist path
Output: certificate (PDF), inline (text summary), or both
Depth: quick (5-item core checklist, ~30s), standard (full checklist, ~2min), deep (full checklist + source re-verification, ~5min)

Workflow

Step 1: Ingest

Read the deliverable. Identify its type (report, script, plan, analysis, email, legal document, other). This determines which checklist items are relevant.

Step 2: Run Agent A (Verifier)

Spawn a sub-agent with these instructions:

You are QC Agent A — a verification specialist. Your job is to check a deliverable against a structured checklist. You must be thorough, skeptical, and evidence-based.

RULES:
- Every finding must include EVIDENCE (command output, source quote, or specific reasoning)
- "I think" is not evidence — verify or mark UNVERIFIED
- Do not assume the deliverable is correct — assume it contains errors until proven otherwise
- Be specific: "Line 14 claims X but source says Y" not "some numbers seem off"

DELIVERABLE:
[Insert deliverable text]

CHECKLIST:
[Insert from references/checklist.md — only items relevant to deliverable type]

For each checklist item, output:
- PASS: [item] — [evidence it's correct]
- FAIL: [item] — [what's wrong + evidence]
- REVIEW: [item] — [unable to verify, reason, suggested human check]

End with a summary: X pass, Y fail, Z review.

Read the full file on GitHub · 178 lines

Files

What ships with it

6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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.

12d ago First seen · 178 lines · 138 tokens per session scan A 4f94fb901dc6

Subscribe to this mod's changes

adversarial-qc is a skill published in the GitHub repository LegalQuants/lq-skills (54 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 138 tokens to every session and 2,853 once invoked, about $0.0007 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 2026-08-30.

Multi-Agent Testing adversarial certificate quality-control verification

adversarial-qc