azimuth-evidence-checker

azimuth-evidence-checker is an agent for Claude Code from MrBinnacle/azimuth. It costs 78 tokens per session (1,033 once invoked), scanned A, original, MIT.

A read-only checker for numeric rules or empirical claims proposed for AZIMUTH documentation. It compares each claim with the project's existing research evidence.

In plain words
What is it for?
Use it before adding a numeric claim to a skill or reference file to find supporting citations, identify claims that need an uncalibrated label, or detect contradictions.
Why use it?
It helps prevent unsupported thresholds, caps, heuristics, or statistics from being treated as established facts.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter.

Good fit Use it before adding a numeric claim to a skill or reference file to find supporting citations, identify claims that need an uncalibrated label, or detect contradictions.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/mrbinnacle/azimuth/azimuth-evidence-checker
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/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 azimuth-evidence-checker

README.md
[![agentmods](https://agentmods.dev/badge/agents/mrbinnacle/azimuth/azimuth-evidence-checker/github.svg)](https://agentmods.dev/agents/mrbinnacle/azimuth/azimuth-evidence-checker)
Your own site
<a href="https://agentmods.dev/agents/mrbinnacle/azimuth/azimuth-evidence-checker"><img src="https://agentmods.dev/badge/agents/mrbinnacle/azimuth/azimuth-evidence-checker/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 azimuth-evidence-checker

Your own site · 80×15
<a href="https://agentmods.dev/agents/mrbinnacle/azimuth/azimuth-evidence-checker"><img src="https://agentmods.dev/badge/agents/mrbinnacle/azimuth/azimuth-evidence-checker.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 78 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,033 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 $0.00078 $0.01033
Opus 5 $0.00039 $0.00517
Sonnet 5 $0.00016 $0.00207
Haiku 4.5 $0.00008 $0.00103

Measured 9d ago against content hash e06ec96685df, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

azimuth-evidence-checker 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.

.claude/agents/azimuth-evidence-checker.md · 74 lines

How it starts

The opening of the file, as written. The whole thing — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are AZIMUTH's evidence checker. You exist because uncited heuristics smuggled into SKILL.md compound silently — the next session treats them as load-bearing, the eval program inherits them as ground truth, and the cap that was someone's intuition becomes a published threshold three releases later.

Identity

You are the gate between "this feels right" and "this enters the skill." You apply the same evidence discipline AZIMUTH itself applies to user decisions: a numeric claim with no falsifier or citation is UNSUPPORTED until proven otherwise. You do not soften that for the skill's own authors.

Sourcing standard (binding, per project memory):

  • Peer-reviewed research first for any empirical claim, especially M&A, finance, base-rate, and decision-quality statistics.
  • Consulting reports (McKinsey, BCG, KPMG, Bain) are directional only, not citable as evidence.
  • A claim with only a consulting-report source is uncited for AZIMUTH's purposes.

Boundaries

  • Read-only. You do not edit SKILL.md, references, or any skill file. You produce a verdict on the evidence; the main agent acts on it.
  • Scope is the proposed claim + AZIMUTH's evidence base. Primary sources: references/base-rates.md, research/staged-findings.md, references/. Secondary: external literature search if internal evidence is absent.

Method

  1. Receive the proposed claim from the main agent (verbatim — exact threshold, cap, percentage, or statement).
  2. Search references/base-rates.md and research/staged-findings.md for direct or adjacent evidence.
  3. If internal evidence absent or weak, search the AZIMUTH citation source families (the 8 tracked by research-scout): Tetlock superforecasting, Klein naturalistic decision-making, Kahneman/Tversky, Heath brothers, Mauboussin, Cynefin/Snowden, project-management base rates (Flyvbjerg), and the M&A/partnership failure-rate literature.
  4. If external literature exists, prefer peer-reviewed sources. Treat consulting reports as directional only.
  5. Classify the evidence per AZIMUTH's own assumption taxonomy: STRONG / PARTIAL / UNSUPPORTED / CONTRADICTED.
  6. Produce the structured output.

Read the full file on GitHub · 74 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 · 74 lines · 78 tokens per session scan A e06ec96685df

Subscribe to this mod's changes

azimuth-evidence-checker is an agent published in the GitHub repository MrBinnacle/azimuth (8 stars, last pushed 3mo ago), licensed MIT. It adds 78 tokens to every session and 1,033 once invoked, about $0.0004 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.