Claire's Appraisal Agent: the free kit

what it does, and what it won't

One lead agent and six specialists. They do the digging and the writing. You make the calls.

It all runs inside your own Claude. No website, hosting or SMS to set up. The private owner page we built around it is a separate build, and optional: it's the full system, at the end.

Before you rely on it, open the listings behind the comparables you lean on. It cites them so you can.

what you need

Six things, and you probably have most of them. The last row is optional.

install it

Do this once. Tick each step off as you go.

That's the chat route, on claude.ai or the Claude desktop app, and it's all you need: in chat, Claude does the six specialists' jobs itself, one after the other (they show greyed out in chat; that's expected). It takes longer, and it does the same job. The plugin you uploaded is saved to your Claude account, so there's no second upload: Cowork (the desktop app's mode where Claude works through a whole job on its own) and Claude Code (Claude's app for developers) pick it up when you're signed in to the same account. In Cowork, open the Cowork tab, then Customize, to see it. There, and in Claude Code, the kit is built to run the three researchers side by side. This kit doesn't need either: the website or the desktop app is enough, and there's no coding. Menus move: if yours look different, search Claude's help for plugins, code execution or web search.

your first message

Copy it, fill in the square brackets and send it in a new chat, on the Claude website or the desktop app. Swap the first line (highlighted) for each job; keep the rest every time. In Cowork (the desktop app's mode where Claude works through a whole job on its own) or Claude Code (Claude's app for developers), you can also type /claire-appraisal-agent:appraise followed by the job. This kit doesn't need either, and there's no coding.

Use Claire's Appraisal Agent.
My agency's details: [agency name], [your name and title], licence [number, if your state needs it], [phone], [email]. Our fees are [your fees from your agreement]. We include [what you include free].
Research it, give me a low, likely and stretch with the evidence, and show what an empty week costs the owner. I'll set the price.

the three rules it runs on

1. The agent researches, a person prices. It gives you the range and the evidence. The figure the owner sees is yours.
2. Show what empty costs. What one empty week costs the owner, next to how long homes like theirs take, so the owner chooses with real numbers.
3. If it's free anyway, show it. Whatever your agency includes at no cost goes in the appraisal, where the owner can see it, instead of in the fine print.

keep it safe

• A person prices and sends. Never send its range as your price. Set the figure yourself, and it writes yours in. The kit never sends anything.
• Check what you lean on. Open the listing behind every comparable that carries your number.
• Your state's rules come first. Estimates and price advertising are regulated, including underquoting for sales, and the rules differ by state. The kit looks up your own state's rules; our two legal notes, in the small print at the very bottom of this page (on paper, the last page of the setup guide), are written for the ACT. When in doubt, ask your licensee or compliance person. This kit is not legal advice.
• Work accounts only. What you paste, like an owner's name and address, goes into your own Claude account, never to us. Use the Claude account on your work email, never one on your personal email.
• Check where chats are kept. Where Claude stores your chats is set out in Anthropic's privacy policy (Anthropic is the company that makes Claude). Check it fits your agency's own privacy policy before you add any owner's details.
• Switch off model training first. It's the Switch off model training step in install it, before any client work: on Pro or Max (and Free), select your name, then Settings, then Privacy, and switch off Help Improve our AI models. While it's on, Anthropic can use your chats to train its models (teach its AI). Anthropic says Team and Enterprise chats aren't used for training by default.
• No thumbs up or down on a client chat. On any plan, Anthropic says that feedback sends it the whole chat, kept for up to five years, and may be used for training even with the setting off.
• No passwords or app keys. Never paste a password, an app key (the secret code one app uses to connect to another) or a bank login into Claude. The kit never needs one.