verdict-auditor

verdict-auditor is a skill for Claude Code from MrBinnacle/azimuth. It costs 57 tokens per session (1,428 once invoked), scanned A, original, MIT.

A diagnostic tool that checks whether an AZIMUTH decision analysis followed AZIMUTH's required structure and rules. AZIMUTH is a method for stress-testing important go-or-no-go decisions.

In plain words
What is it for?
Use it after an AZIMUTH session to check the opening verdict, section content, and other required rules, with each issue rated by severity.
Why use it?
It finds structural mistakes and quality drift without rewriting the analysis, so problems can be fixed separately.

Skill for Claude Code

Written for Claude Code: installed under .claude/.

Good fit Use it after an AZIMUTH session to check the opening verdict, section content, and other required rules, with each issue rated by severity.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/mrbinnacle/azimuth/verdict-auditor
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 MrBinnacle/azimuth --skill verdict-auditor
Clone the repo
git clone --depth 1 https://github.com/MrBinnacle/azimuth

Made for: Claude Code.

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 verdict-auditor

README.md
[![agentmods](https://agentmods.dev/badge/skills/mrbinnacle/azimuth/verdict-auditor.svg)](https://agentmods.dev/skills/mrbinnacle/azimuth/verdict-auditor)
Your own site
<a href="https://agentmods.dev/skills/mrbinnacle/azimuth/verdict-auditor"><img src="https://agentmods.dev/badge/skills/mrbinnacle/azimuth/verdict-auditor.svg" alt="Measured on agentmods" height="20"></a>
Per session 57 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,428 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 $0.00057 $0.01428
Opus 5 $0.00028 $0.00714
Sonnet 5 $0.00011 $0.00286
Haiku 4.5 $0.00006 $0.00143

Measured 8d ago against content hash dcd586beb32f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

verdict-auditor 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 8d 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.

.claude/skills/maintenance/verdict-auditor/SKILL.md · 156 lines

How it starts

The opening of the file, as written. The whole thing — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Verdict Auditor

Apply AZIMUTH's own rules against its own output. Find where the skill drifted from its discipline.

This skill does not improve the output. It identifies exactly what rule was violated and how severely. Rewrites happen in a separate session.


Input

User pastes an AZIMUTH output. If none is provided, ask: "Paste the AZIMUTH output to audit."


Audit Checks

Run all 8 checks. Assign each finding a severity: CRITICAL / MODERATE / MINOR.


Check 1 — Verdict-First Rule

Rule: The first three elements of every output must be: (1) Azimuth Verdict line, (2) Recommended Decision, (3) Confidence Level. Nothing else precedes them.

How to check: Read the first 10 lines. Does anything appear before the verdict line?

Severity if violated: CRITICAL — the reader cannot act on the first paragraph alone.


Check 2 — Empty Section Rule

Rule: Do not emit a section header with no substantive content. If a section has nothing genuine, it must be omitted entirely.

How to check: Scan all section headers. Is there any header followed by a placeholder, "N/A," "none identified," or a single generic sentence?

Severity if violated: MODERATE — signals padding; reduces trust in sections that do have content.


Check 3 — Register Deduplication

Rule: Modules 2, 5, 6, 7, and 8 feed a shared register. The same risk or assumption must not appear independently in multiple output sections. It is cited once in the most relevant section; other sections reference it by tag.

How to check: Does the same underlying risk appear in both Critical Risks and Likely Failure Paths as if they are independent? Does the same assumption appear in both Weak Assumptions and Critical Risks as a separate entry?

Severity if violated: MODERATE — register duplication bloats output and obscures which entries are actually high-severity.


Check 4 — Failure Path Source Check

Rule: Likely Failure Paths reuse register entries. They do not introduce risks that are not already in Critical Risks.

Read the full file on GitHub · 156 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. 8d ago First seen · 156 lines · 57 tokens per session scan A dcd586beb32f

Subscribe to this mod's changes

verdict-auditor is a skill published in the GitHub repository MrBinnacle/azimuth (8 stars, last pushed 3mo ago), licensed MIT. It adds 57 tokens to every session and 1,428 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.