Chat Assistant Disclosure Desk: safe chat assistant rules and a ten-check chat scan (free sample)
by PaperChaseWebb
From Chat Assistant Disclosure Desk: the ways website chat assistants slip, ten rules to paste into your agent, and a read-only script that runs ten checks on one saved chat, from no AI line in the first message and "yes, I'm a real person" to a request for a person with no contact given, made-up waivers, card number requests and bookings that never happened.
- Price
- Free
- Difficulty
- Beginner
- Setup
- 5 min
- Format
- .md file
chat-assistant-disclosure-desk-safe-chat-assistant-rules-ten-check-chat-scan-free-sample.md · 11 KB · direct download, no signup
What you end up with
You can see whether a saved chat said it was AI, answered honestly and handed off properly before you trust the assistant with more visitors.
What you get
- Why chat assistants need checking
- Ten rules for agents that write or tidy chat assistants
- chat_mini.py: ten read-only checks for one saved chat
- A made-up water heater chat with seven findings to try it on
Who it is for
Anyone who wants a quick check on what their website chat assistant actually told a visitor.
Works with
Solid chips: the agent is named in this listing's own tools list. Dashed chip: Plain Markdown, so it also pastes into other agents such as Grok Bot, Hermes, OpenClaw or Muse. We list those as compatible by format only, not as tested.
Tools needed
Cost to run: Free.
Time to result: About five minutes to a first scan
Sample prompts, free to use
Run python3 chat_mini.py on each chat in [folder]. List every finding by chat, and tell me which ones need a change to the assistant's greeting or instructions. Do not change the live chat.
The rules and script are in the file.
Free preview of the file
# Chat Assistant Disclosure Desk: safe chat assistant rules and a ten-check chat scan (free sample) Version 1.0, 2026-11-01. By PaperChaseWebb. For anyone with an AI chat assistant on their website, or letting an AI agent tidy one up. One page and one Python 3 script, standard library only, read-only: it reads one saved chat, prints findings and changes nothing. It sends nothing and opens no network connection. No API key, no login. Tested on Linux with Python 3.13, 3.12, 3.11, 3.9 and 3.8: 3 unit tests, and 21 of 21 planted faults caught. The example business is made up. Not legal advice. ## Why chat assistants need checking Maine requires a business chatbot to tell consumers clearly they are not talking with a human. California requires bots that sell to say so, and from 2027 its largest businesses must get a person to a customer who asks. The EU requires people to be told they are talking to AI. And when Air Canada's chatbot made up a refund policy, a tribunal held the airline to it. Most small business chat assistants slip in the same few ways: - The greeting says "Hi, I'm Jess!" and never says it is AI. - Asked "are you a real person?", it says yes. - Asked for a person, it says "No need for a person, I can help you faster!" - It quotes a price, waives a fee or promises a warranty nobody wrote down. - It says "I've booked you for Tuesday" when it cannot book anything. - It asks for a card number in chat. ## Rules to paste into AGENTS.md, CLAUDE.md, a Grok Bot stand
Test record
- Date
- Nov 1, 2026
- What was run
- tests/test_mini.py (3), 21 planted faults against chat_mini.py, and the copy extracted from the sample page. Linux, Python 3.13, 3.12, 3.11, 3.9 and 3.8.
- Result
- 3 passed, 0 failed on all five versions; 21 of 21 faults caught; the extracted copy was identical to the tested script and found 7 problems on the example chat.
- Not run
- Your settings and facts, the assistant file and page checks, the fix check and the eval set (the full kit does those).
Copied from this item's dated test notes. What was not run stays in.
Worth knowing
- The sample checks one chat on its own. Your settings and facts, the assistant file and AI use page checks, the fix check against the original, the eval set, owner approval and agent setups are in the full kit.
- Not legal advice. It does not say a chat meets any law.
- On 2026-11-01 chat_mini.py passed 3 unit tests and 21 of 21 planted faults on Linux with Python 3.13, 3.12, 3.11, 3.9 and 3.8, and the copy extracted from the sample page was identical to the tested script.
Questions
Does it change my files or send anything?
No. It reads the file you name and prints findings. It opens no network connection, sends nothing and changes nothing.
What is in the full kit?
19 checks on the greeting, instructions and handoff text, 12 on your AI use page, 16 on every saved chat, a 12-check fix check against the original, owner approval at a terminal, a 10-chat eval set, and setup for Grok Bot, Claude Code, Codex and Hermes. $29.
Details
Version 1.0 · Updated Nov 1, 2026 · Markdown version of this page
License: Personal and commercial use. Download terms
Related
Chat Assistant Disclosure Desk: agent fixes for your website chat, AI said up front, a person on request, owner approved
Let an agent tidy your website chat assistant's greeting, instructions, handoff text and AI use page, and review a week of saved chats, without the assistant passing as a person, keeping visitors from a person, or making up prices and promises. It writes fixed copies and never touches the live chat. A fix check refuses a removed AI line, a deleted emergency line, a new discount or a new phone number. A chat check flags every reply that claimed to be human, ignored a request for a person or quoted a price you never wrote down. You approve at your terminal and paste it in yourself.
Text Message Consent Desk: safe text fixing rules and a ten-check text scan (free sample)
From Text Message Consent Desk: the mistakes agents make when they tidy a business's texts, nine rules to paste into your agent, and a read-only script that runs ten checks on one saved text, from a missing STOP line and opt-outs that ask for a log-in to link shorteners, fake Re: lines, offers inside reminders, late send times and compliance claims.
Review Reply Desk: safe reply rules and a ten-check review reply scan (free sample)
From Review Reply Desk: why review replies need checking, eight rules to paste into your agent, and a read-only script that runs ten checks on a reply draft, from public offers and requests to change a review to last names, invented numbers and links you did not approve.