hallucination-audit

hallucination-audit is a command for coding agents from bitflight-devops/hallucination-detector. It costs 29 tokens per session (453 once invoked), scanned A, original, MIT.

A command for checking text for speculation, unsupported cause-and-effect claims, invented measurements, and unsupported claims of completeness.

In plain words
What is it for?
Use it after drafting an answer or reviewing output from another agent.
Why use it?
It helps identify statements that sound certain but are not backed by observed evidence.

Command

Part of the hallucination-detector plugin — 4 skills, 3 commands, 5 agents, 4 hooks, 1 MCP server shipped together

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/bitflight-devops/hallucination-detector/hallucination-audit
Clone the repo
git clone --depth 1 https://github.com/bitflight-devops/hallucination-detector

Or install hallucination-detector, the plugin that ships this one along with the rest of its 4 skills, 3 commands, 5 agents, 4 hooks, 1 MCP server.

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 hallucination-audit

README.md
[![agentmods](https://agentmods.dev/badge/commands/bitflight-devops/hallucination-detector/hallucination-audit.svg)](https://agentmods.dev/commands/bitflight-devops/hallucination-detector/hallucination-audit)
Your own site
<a href="https://agentmods.dev/commands/bitflight-devops/hallucination-detector/hallucination-audit"><img src="https://agentmods.dev/badge/commands/bitflight-devops/hallucination-detector/hallucination-audit.svg" alt="Measured on agentmods" height="20"></a>
Per session 29 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 453 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 $0.00029 $0.00453
Opus 5 $0.00015 $0.00227
Sonnet 5 $0.00006 $0.00091
Haiku 4.5 $0.00003 $0.00045

Measured 4d ago against content hash 39ecd204ba99, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

hallucination-audit 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.

commands/hallucination-audit.md · 53 lines

What it actually says

Hallucination Audit

INPUT: Paste the target content to audit (assistant output, sub-agent output, or a draft response).

Review the target content for these HALLUCINATION TRIGGERS:

1. Speculation Language (Guessing Trigger)

Scan for: "I think", "likely", "probably", "seems", "should be", "assume", "maybe", "might".

Action: FLAG. Replace with one of:

  • "I don't know yet."
  • "I don't have that information."
  • "This is something I can check using my tools."
  • "I did these steps: . I observed: ."

2. Causality Without Evidence (Diagnosis Trigger)

Scan for: "because", "due to", "caused by", "therefore", "this means", "as a result".

Action: If the sentence does not cite a specific observation (tool output, logs, file lines, measured behavior), FLAG. Rewrite as:

  • Observation-only statement, OR
  • Explicit hypothesis + next verification step (no causal claim yet).

3. Pseudo-Quantification (Fake Rigor Trigger)

Scan for: scores/percentages like "8.5/10", "70% improvement", "100%".

Action: If no methodology + evidence is shown, FLAG. Replace with measurable evidence or remove.

4. Completeness Claims (Overreach Trigger)

Scan for: "all files checked", "comprehensive analysis", "fully resolved", "everything fixed".

Action: FLAG unless the content explicitly lists the concrete checks performed (what was inspected, with what scope).

5. Delegation Micromanagement (Prescription Trigger)

Scan for: overly specific prescriptive edits in delegation prompts ("change line 42", "rename variable X to Y") when not required by the user.

Action: Replace with success criteria + constraints + verification steps.

OUTPUT:

  • Pass/Fail.
  • If Fail: quote the exact triggering phrase(s) and provide the required rewrite.
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 · 53 lines · 29 tokens per session scan A 39ecd204ba99

Subscribe to this mod's changes

hallucination-audit is a command published in the GitHub repository bitflight-devops/hallucination-detector (7 stars, last pushed 1mo ago), licensed MIT. It adds 29 tokens to every session and 453 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.

Related

Other commands, from other repositories