The PracticeThe Practice The PracticeOperating infrastructure

The Marketplace·The Agent Prompt Library

The Agent Prompt Library

When you improvise a new prompt each time: fifty monthly agent jobs, pre-built; run through the library, each draft gets a fair-housing language screen.

What it is

Fifty prompts for the work a Georgia residential agent writes every month, with a routing layer. You type a plain request. The library picks the prompt, asks for what it needs in one message, drafts, and checks its own draft before handing it over.

Rung

Library, a complete prompt set for one practice.

Price

$636, or one category at $106.

Each product is sold as the version delivered. We do not promise updates, fixes or support after delivery. When we release a new major version of a product, half of what you paid for the version you hold counts toward the new one, with no time limit.

Request early access

Or get the Letters, twice a month by email.

Format

76 files, among them:

  • 50 prompts in 9 category folders;
  • a Field Guide (PDF) with install chapters for both platforms;
  • a routing skill for Claude;
  • a Librarian build for ChatGPT, with 11 knowledge files;
  • an index and a quality rubric;
  • a voice-calibration worksheet;
  • a disclaimer.
Who it fits

Licensed Georgia residential agents, solo or on a team, who draft listing copy, client emails, market notes and follow-ups every month.

Version

Field Guide 1.0.7, October 2026.

Requirements

Claude: Any plan, with code execution turned on: skills need it. Upload the routing skill, then add 11 files to one Claude Project. About 10 minutes. Claude Free holds up to 5 Projects; Claude Pro is $20 a month, or $17 a month billed annually. (claude.com/pricing and the Claude Help Center, as of October 29, 2026.) ChatGPT: A Project named Agent Prompt Librarian, holding the Librarian instructions and 11 files. About 15 minutes. Eleven files need Go or above: ChatGPT Go ($8 a month) and Plus ($20 a month) hold up to 25 files per Project (chatgpt.com/pricing and OpenAI's help center, as of October 29, 2026). On the Free plan, copy a prompt from its file, fill in the blanks and paste it into a chat. A prompt copied by hand carries only its own checks, without the fair-housing screen. Optional: The voice-calibration worksheet, about 15 minutes, so every draft sounds like you. You install it yourself. Questions about your purchase: jake@runthepractice.com.

Built and tested on Claude and ChatGPT. Everything is plain files you keep, so you can use it in other assistants. We have not tested those.

What's included

  1. Listing creation, 8 prompts:
    • luxury, mid-market and entry-level descriptions;
    • coming soon; just listed; open house; price reduction;
    • a listing social post.
  2. Buyer side, 7 prompts, from the consultation script to the post-tour follow-up.
  3. Listing presentation, 5 prompts, among them the CMA narrative and the price-adjustment conversation.
  4. Market intelligence, 5 prompts, from the weekly update to a neighborhood deep dive.
  5. Sphere and database, 6 prompts.
  6. Negotiation, 5 prompts, among them appraisal-gap strategy and repair request and response.
  7. Closing and post-close, 5 prompts, among them closing-day instructions and the one-year follow-up.
  8. Prospecting and lead generation, 5 prompts.
  9. Personal brand, 4 prompts, among them your agent bio.
  10. Routing: the Claude skill and the ChatGPT Librarian match a casual request to the best prompt. A close call offers the top two; no match falls back to general drafting and says so.
  11. A verification step inside every prompt. On Claude and in the Librarian, every draft also gets a fair-housing language screen that flags wording for you and your broker to review. A prompt copied by hand carries only its own checks.
  12. A quality rubric, and a template for writing your own prompts in the same format.
  13. A voice-calibration worksheet.
  14. The Field Guide (PDF), with install chapters and acceptance tests for Claude and ChatGPT.

Sample output

Sample: The built-in example in the mid-market listing prompt, on a sample property.

Input, abbreviated as in the prompt's own example: 3-bed, 2.5-bath ranch in Smyrna, $625K, recently updated kitchen and primary bath, fenced yard, finished basement. In use, the prompt asks for every fact the draft states (address, square footage, key features, condition) before it drafts.

"This 3-bedroom Smyrna ranch went through a full kitchen and primary-bath renovation in 2024 and shows like a 2024 build at 1985 prices. The kitchen now has quartz counters, soft-close cabinets, and a peninsula sized for four. Primary bath was opened up to a true walk-in shower with a frameless glass door.

2,200 SF on the main level plus a finished basement with a separate entrance, a wet bar, and a fourth flex room that has worked as an office, a guest space, or a rec room depending on who's using it.

The fenced backyard is level, sodded, and faces west -- afternoon sun for the patio, morning shade for the dogs. Walking distance to the Silver Comet trailhead.

Available for showing now. Call Sarah at the number below to schedule."

The install test: Ask for "a listing description perfect for a young family near great schools." A correct install routes to the listing prompt, raises the buyer-preference and school wording as fair-housing wording, and offers a rewrite that describes the property.

What it won't do

  • It does not decide whether wording is lawful. The screen flags language for you and your broker to review.
  • It does not look up facts. Each prompt asks for the facts it needs (address, price, features, figures), and you supply them.
  • It sends nothing. Every draft is yours to read, edit and send.
  • It is not a separate reviewer of finished drafts. Fair Housing Sweep and Ethics Check add two reviewers that read what you publish.
  • It does not replace your brokerage's forms or its process for agreements. Prompts that touch the Buyer Brokerage Agreement send questions about it to your broker.

Skip it if you need only one kind of job; that category's pack is $106.

Trade-offs

One category packThe Agent Prompt LibraryThe Complete Agent Stack
Price$106$636$954 ($1,272 bought separately)
What you getThe prompts for one kind of job, without routingAll 50 prompts, with routingThis library, Fair Housing Sweep and Ethics Check, The Instruction Framework (Georgia Agent Edition) and a platform setup guide
What you give upThe other eight categories, the routing and its fair-housing screenThe two reviewers and the working rulesNothing in this product; $318 more

Every rung, side by side: What it costs.

Common mistakes

  • Copying prompt text from the PDF page. Copy it from the prompt files. The PDF is for reading.
  • A skill upload that will not take the file. If the upload dialog accepts only .zip files, rename the .skill file to .zip. They are the same archive.
  • Code execution left off on Claude. Skills need it. Turn it on in Settings, under Capabilities, before you upload.
  • Pasting the Librarian instructions only in part. If the install test simply writes the copy without raising the school wording, the instructions did not paste completely.

Georgia notes

  • Written for licensed Georgia residential agents. The worked examples are set in metro Atlanta markets such as Smyrna and East Cobb.
  • Buyer-side prompts use the Buyer Brokerage Agreement by name, and they send questions about it to your broker.
  • The offer cover letter prompt adds a note for you: some brokerages and listing agents discourage cover letters, or will not forward them, because of fair-housing concerns.

Questions

How is this different from writing my own prompts?

Each prompt carries its required inputs, its constraints, a slot for your voice and a verification step, and the routing picks the right one from a plain request. You can still write your own; a template in the library keeps them in the same format.

Does it run on ChatGPT?

Yes, as a ChatGPT Project called the Agent Prompt Librarian, with the same 50 prompts and the same verification step. On Claude the routing works on its own; on ChatGPT you open the Librarian to use it.

Is the ChatGPT edition a custom GPT?

No. OpenAI is retiring custom GPTs (Help Center FAQ, checked October 29, 2026), so the ChatGPT edition installs as a Project.

Does the fair-housing screen tell me whether my listing is legal?

No. It flags wording for you and your broker to review and offers a rewrite that describes the property. Decisions about wording stay with you and your broker.

Can I edit the prompts?

Yes. Every prompt is a plain file you can open and change, under a license for your own work. What you write with them is yours.

Do I need a paid AI plan?

On Claude, any plan works with code execution turned on. On ChatGPT, the Project install holds 11 files, so it needs Go or above; on the Free plan you copy prompts from the files by hand, and each carries only its own checks.

Can my team or brokerage use one copy?

No. The license is personal to you and cannot be shared or deployed for others in your organization. A team or a brokerage that wants it for several people should ask us about that.

What is the refund policy?

A full refund within 14 days of purchase, no questions asked, once per product. Write to jake@runthepractice.com. A refund ends your license to that product.

Fine print

Prices are in U.S. dollars. Sold by Stoa LLC. Payments are processed by Stripe. Sales tax is calculated at checkout where it applies. Available to customers in the United States only. A purchase is a license for your own professional use, not ownership, and what you write with it is yours. One-time purchase. No subscription. These are tools, not legal advice, and buying them makes you no one's client.