verify-claims

verify-claims is a command for Claude Code from basher83/agent-auditor. It costs 11 tokens per session (1,248 once invoked), scanned A, original, MIT.

A command for checking factual statements in documentation against the actual code and Git history. It examines what the code does rather than trusting filenames or descriptions.

In plain words
What is it for?
Use it to verify statements about implementations, features, tests, or pull requests, using repository evidence for each claim.
Why use it?
It helps catch documentation claims that are unsupported, outdated, or contradicted by the repository.

Command for Claude Code

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.

agentmods
npx agentmods add commands/basher83/agent-auditor/verify-claims
Clone the repo
git clone --depth 1 https://github.com/basher83/agent-auditor

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 verify-claims

README.md
[![agentmods](https://agentmods.dev/badge/commands/basher83/agent-auditor/verify-claims.svg)](https://agentmods.dev/commands/basher83/agent-auditor/verify-claims)
Your own site
<a href="https://agentmods.dev/commands/basher83/agent-auditor/verify-claims"><img src="https://agentmods.dev/badge/commands/basher83/agent-auditor/verify-claims.svg" alt="Measured on agentmods" height="20"></a>
Per session 11 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,248 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00011 $0.01248
Opus 5 $0.00005 $0.00624
Sonnet 5 $0.00002 $0.00250
Haiku 4.5 $0.00001 $0.00125

Measured 4d ago against content hash 316c9b9169d3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

verify-claims 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 4d 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/commands/verify-claims.md · 204 lines

How it starts

The opening of the file, as written. The whole thing — 204 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Task: Verify Claims About Code

Document to verify: $ARGUMENTS

Purpose: Verify factual claims in documents about code implementations, features, tests, or PRs.

Core 4 Foundation:

  • Context: Document with claims + git repository state
  • Model: Semantic verification capabilities
  • Prompt: This systematic verification process
  • Tools: Git commands, file operations

Instructions

You are verifying claims in the document specified above. Your job is to check facts against source truth, not provide philosophical insights.

Core Principles

  1. Evidence before claims: Never mark verified without git command output proving it
  2. Verify behavior, not names: Check what code DOES. Different filenames can provide same functionality.
  3. Counter-evidence first: Define what would disprove claim BEFORE checking
  4. Cross-reference patterns: If "X missing Y but Z has Y," check what Y IS in Z first
  5. No meta-analysis: No recursive patterns, cognitive dynamics, or turtles

Verification Process

For each claim:

  1. Extract claim: Quote exactly
  2. Identify verification need: What file/code would prove/disprove?
  3. Define counter-evidence: "This claim would be FALSE if I found [specific evidence]"
    • Focus on FUNCTIONALITY not filenames
  4. Cross-reference if comparing: Check what feature IS in reference first
  5. Run verification: Use git show/diff/grep
  6. Show evidence: Paste relevant output
  7. State verdict: ✅ Verified / ❌ Incorrect / ⚠️ Partially true
  8. Move to next claim: No commentary, just next verification

Example: The Naming vs Functionality Pattern

Claim: "PR #7 missing Jina Reader direct API variation"

What this means: PR #7 lacks script making direct HTTP calls to r.jina.ai (not using MCP)

Counter-evidence: ANY script in PR #7 that:

  • Uses httpx/requests (not MCP dependencies)
  • Calls https://r.jina.ai/
  • Downloads docs directly

Commands:

# Check what files exist
git diff --name-only main...feature-branch

# Check dependencies
git show feature-branch:path/to/script.py | head -20

# Check for API calls
git show feature-branch:path/to/script.py | grep "r.jina.ai"

Read the full file on GitHub · 204 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. 4d ago First seen · 204 lines · 11 tokens per session scan A 316c9b9169d3

Subscribe to this mod's changes

verify-claims is a command published in the GitHub repository basher83/agent-auditor (5 stars, last pushed 6mo ago), licensed MIT. It adds 11 tokens to every session and 1,248 once invoked, about $0.0001 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.