A listing description is about 150 words. It is also a regulated advertisement. One lazy phrase like “perfect for young families” can turn a routine MLS entry into a Fair Housing complaint, and the agent’s name is on it, not the AI’s.
That is why real estate marketing is a bad fit for a blank chat window and a good fit for Claude Code skills. A skill is a folder with a SKILL.md file that gives Claude standing instructions for one job. Write the compliance rules once, and every listing, neighbourhood page and nurture email gets checked against them without you pasting a rules list into every prompt.
This guide covers the skills worth installing for four jobs: listing descriptions, neighbourhood pages, local SEO, and agent email nurture. You get the source, licence, GitHub stars and last push (checked through the GitHub API on 2 November 2026), install commands, example prompts, and what the output should contain. It also covers the official Fair Housing sources, because a skill is only as good as the rules it was written from.
One honest note before we start: this post is research-based. We read the repos, the skill files and the official sources. We did not run a controlled test of every skill on real listings, and we say so wherever it matters.
New to skills? Read what Claude Code skills are first.
The short answer: which Claude Code skills should real estate marketers install?
Start with the free Prosperity Agent fair-housing-overlay and listing-description skills for MLS copy, add Corey Haines’ marketingskills for email and page copy, and add claude-seo if you build neighbourhood or city pages that need to rank. All of these are free and MIT licensed. Then write your own short compliance file based on HUD’s rules and your state’s protected classes, because no public skill fully covers your state.
Paid packs exist, but none we found is built for real estate. They are general marketing bundles you would still need to adapt.
The shortlist at a glance
| # | Skill or pack | Source | Free or paid | Best for | Stars (2 Nov 2026) | Last push |
|---|---|---|---|---|---|---|
| 1 | prosperity-real-estate-skills | github.com/TheProsperityAgent/prosperity-real-estate-skills | Free (MIT) | Listing copy, Fair Housing overlay, MLS data | 11 | 26 Jul 2026 |
| 2 | mls-listing-craft | claudskills.com (source: tinh2/skills-hub-registry) | Free skill; optional Pro $9/mo | Five listing lengths plus RealEstateListing JSON-LD | n/a | n/a |
| 3 | marketingskills | github.com/coreyhaines31/marketingskills | Free (MIT) | Email sequences, landing pages, copy | ~57.9k | 31 Oct 2026 |
| 4 | claude-seo | github.com/AgriciDaniel/claude-seo | Free (MIT) | Local SEO, schema, page audits | ~25.6k | 1 Nov 2026 |
| 5 | Anthropic Marketing plugin | github.com/anthropics/knowledge-work-plugins | Free (Apache-2.0) | Campaign plans, brand voice review | ~34.4k (repo) | 1 Nov 2026 |
| 6 | anthropics/skills (docx, pdf, pptx, xlsx) | github.com/anthropics/skills | Free to use, source-available, not open source | Flyers, listing packs, CMA decks | ~186k | 28 Oct 2026 |
| 7 | tech-in-real-estate | github.com/analyticsariel/tech-in-real-estate | Free (MIT) | Investor-side skills, agents and MCP servers | 2 | 20 Apr 2026 |
| 8 | Inflectual Master Marketing Skills Bundle | inflectual.gumroad.com | Paid: $167 one-time | Broad marketing library | n/a | n/a |
| 9 | Contra Ultimate Claude SEO and AEO Skills Pack | contra.com | Paid: $39 (shown against $79) | SEO and AI-answer content | n/a | n/a |
| 10 | SuperSkills | superskills-one.vercel.app | Paid: $50 | General skill pack | n/a | n/a |
Why real estate marketing needs skills, not just prompts
Real estate copy is advertising under federal law, so the same rules apply to every piece you publish. Skills are where you store those rules once, so Claude applies them every time without you remembering to ask. A prompt is a one-off. A skill is standing policy.
Here is the problem in plain terms. Section 804(c) of the Fair Housing Act makes it unlawful to make, print or publish any housing advertisement that indicates a preference, limitation or discrimination based on race, colour, religion, sex, handicap, familial status or national origin. HUD’s regulation at 24 CFR 100.75 applies that to written and spoken statements, including advertisements. Your state may add classes such as source of income, sexual orientation or age.
A general model writes from patterns in the text it learned from. Plenty of old listing copy says “great schools”, “family neighbourhood” or “ideal for empty nesters”. Ask a model for a warm listing description with no guardrails, and some of that language can come straight back.
What a skill changes
| Without a skill | With a compliance skill loaded |
|---|---|
| You paste rules into each prompt, or forget | Rules load automatically when the task matches |
| Each agent on a team uses different wording | One file sets the house standard |
| Rule changes mean retraining everyone | You edit one SKILL.md |
| No record of what the model was told | The skill file is a reviewable record |
The last row matters more than it looks. If a complaint ever lands, being able to show your brokerage the written rules your AI drafts were checked against is better than saying “I asked it to be careful.”
Two limits apply. A skill reduces risk; it does not remove it. And a skill’s rules are only as accurate as the person who wrote them, which is why the next section goes to the primary sources.
Fair Housing rules your skills must follow (with the HUD sources)
The primary sources are the Fair Housing Act section 804(c) (42 U.S.C. 3604(c)), HUD’s regulation at 24 CFR 100.75, and HUD’s 9 January 1995 memorandum on advertisements under section 804(c), signed by Assistant Secretary Roberta Achtenberg. Build your compliance skill from these, not from a blog list of banned words. One trap: many skills and blogs still cite “24 CFR part 109” as HUD’s advertising rules. HUD removed part 109 from the Code of Federal Regulations in 1996 (61 FR 14378), so it is old, non-binding guidance at best. Cite 100.75 and the 1995 memo instead.
What the 1995 HUD memo says
The 1995 memo is the document most “banned words” lists are loosely based on, and it is more permissive than many of those lists. Based on the memo text as quoted by Texas A&M’s Real Estate Center and the Oregon Realtors association:
- Ads that describe the property (“two bedroom”, “cozy”, “family room”), services and facilities, or the neighbourhood (“quiet streets”) are not treated as facially discriminatory.
- The memo says complaints over phrases such as “master bedroom”, “rare find” or “desirable neighborhood” should not be filed.
- Describing accessibility features of the property is allowed. A phrase like “walking distance to bus stop” was not treated as a violation.
- Statements of preference or limitation tied to a protected class, such as “adults only” or “no children” for housing that does not qualify for the senior housing exemption, are the core problem.
Context still matters. The Oregon Realtors article notes that some associations encourage “primary bedroom” anyway, and NAR training materials stress that how a phrase is used can change its meaning.
The 2024 digital advertising guidance was withdrawn
In April 2024, HUD’s Office of Fair Housing and Equal Opportunity issued guidance on advertising through digital platforms, including ad targeting. A Federal Register notice of withdrawal of FHEO guidance documents (docket FR-6571-N-01, published 6 April 2026) lists that guidance as withdrawn, effective 17 September 2025. The statute and 24 CFR 100.75 are unchanged, so the advertising rules still apply. If a skill or blog cites the 2024 guidance as current, treat it as out of date.
What to put in your compliance skill
| Rule | Why | Source |
|---|---|---|
| Describe the property, not the buyer | 804(c) targets preference for kinds of people | 42 U.S.C. 3604(c) |
| No statements about who “should” live there | Steering by familial status, religion, etc. | 24 CFR 100.75 |
| Neutral property terms are fine | HUD says facially neutral terms do not create liability | 1995 HUD memo |
| Add your state’s protected classes | State laws can be broader than federal | Your state civil rights agency |
| Run paid ad targeting past your broker | Ad delivery can be a separate issue from copy | FHA; check current HUD notices |
| Include the Equal Housing Opportunity logo or statement where your brokerage requires it | HUD logo practice (former part 109) and brokerage policy | Broker policy |
This is not legal advice. Your broker, your MLS and a fair housing attorney in your state are the final word.
Best free Claude Code skills for real estate marketing
The best free real estate skill set we found is TheProsperityAgent’s prosperity-real-estate-skills, which is small (four skills), MIT licensed and built by licensed agents. Pair it with general marketing and SEO packs for everything beyond listing copy. Real-estate-specific repos are young and have few stars, so read every file before you trust it.
1. prosperity-real-estate-skills (TheProsperityAgent)
Built by Al and Victoria Pinder, who state in the README that they are licensed agents with eXp Realty in North Carolina. Four skills:
listing-description: turns listing facts into MLS remarks plus a short social caption.connect-mls-to-claude: a guide to loading your MLS and county exports so Claude works from real numbers.fair-housing-overlay: a compliance layer meant to load alongside every other skill.quick-start-3-wins: three starter workflows.
Install as a plugin:
/plugin marketplace add TheProsperityAgent/prosperity-real-estate-skills
/plugin install real-estate-skills@prosperity-real-estate-skillsThe README is upfront about the business model: the skills are free, and the authors earn revenue share if you join eXp Realty and name them as your sponsor. That disclosure is a point in its favour. The README also shows a sample output and says a hand-written “before” version had eleven compliance issues. That is the authors’ own example, not an independent test.
Read the overlay before you rely on it. It bans school references outright, then lists “school-distance-in-miles” as a suggested substitute a few lines later. It also states that clichés like “must-see” are filtered out by Realtor.com, Zillow and Redfin. We found no source for that claim. Fix both in your fork.
2. mls-listing-craft (ClaudSkills registry)
Listed on claudskills.com from the tinh2/skills-hub-registry repo. According to the listing, it produces five variants (tagline, MLS short up to 250 characters, MLS long up to 4,000 characters, a 600 to 800 word marketing description, and a social caption) plus schema.org RealEstateListing JSON-LD, and runs a Fair Housing check. Install by copying SKILL.md into ~/.claude/skills/mls-listing-craft/. The skill is free; the registry sells an optional Pro tier at $9 a month, $79 a year or $149 one-time. Check your MLS character limits, which vary by board.
3. marketingskills (Corey Haines)
Not real-estate-specific, but its email sequence, copywriting and page CRO skills handle nurture emails and landing pages well once you add your compliance file. See our marketingskills review.
npx skills add coreyhaines31/marketingskills4. claude-seo (AgriciDaniel)
Local SEO audits, schema generation and page-level checks. Useful for neighbourhood and city pages. Covered in depth in Claude Code skills for local SEO.
5. Anthropic Marketing plugin
Campaign planning, brand voice review and performance reports. Install:
claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install marketing@knowledge-work-plugins6. anthropics/skills document skills
The docx, pdf, pptx and xlsx skills build listing packs, open house flyers and CMA decks. They are source-available, not open source, so read the licence before redistributing modified versions.
7. tech-in-real-estate (analyticsariel)
MIT licensed, aimed at investors, with skills, Python agents and MCP servers. Small (2 stars) and last pushed April 2026. Worth a look for deal analysis, less so for agent marketing.
Paid Claude Code skill packs: worth it for agents?
None of the paid packs we verified is built for real estate, so for most agents the free stack above plus your own compliance skill beats a paid bundle. Paid packs make sense if you also run SEO content or ads at volume and want a wide library on day one.
| Pack | Price (verified) | What you get | Real estate fit |
|---|---|---|---|
| Inflectual Master Marketing Skills Bundle | $167 one-time | 167 marketing skills | General; add your own Fair Housing rules |
| Contra Ultimate Claude SEO and AEO Skills Pack | $39 (listed against $79) | SEO and AI-answer content skills | Fits neighbourhood pages |
| SuperSkills | $50 | General skill pack | General |
| Real Estate Agent Toolkit (parclarktech, Gumroad) | Pricing not public on the page we checked | 9 “AI skills”: listing copy, CMA narrative, follow-up, farming letters | Real estate, but the page does not say it is built for Claude |
Prices were taken from each vendor’s page or our verified list on 2 November 2026. Check them before buying, since Gumroad and Contra prices change. On the parclarktech toolkit: the page describes “9 AI skills” with a fair-housing filter, but does not mention Claude or Anthropic, so we cannot confirm it installs as Claude Code skills.
You also need a Claude plan. Claude Pro is $20 a month, or $17 a month billed yearly, and includes Claude Code.
Workflow 1: listing descriptions that pass a Fair Housing check
Feed Claude verified facts only, load your compliance skill, ask for every length your channels need, and then run a separate review pass before a human approves it. Generation and review should be two steps, never one.
The input
Give Claude a structured fact sheet, not a vibe. Pull it from the MLS input form or your listing appointment notes: beds, baths, square footage, lot size, year built, updates with dates, features by room, HOA facts, and distances to named, non-protected places (a station, a highway, a park).
Example prompt
Use listing-description with fair-housing-overlay.
Facts: see listing-123.md. Only use facts in that file.
Write: MLS public remarks (max 1,000 characters), a 60-word
portal teaser, an Instagram caption, and a 20-word headline.
Then list every claim you made and the line in the fact file
that supports it.What the output should contain
- Each version within your MLS and portal character limits.
- Property features only. No sentences about who would enjoy the home.
- No school quality claims. Many brokerages also avoid naming schools at all; follow your broker.
- A claim-to-fact table, so you can spot invented features (a “chef’s kitchen” that is a galley, an “updated roof” with no date).
- A flags list: any phrase the review pass thinks is borderline, with a neutral alternative.
The review pass
Run a second prompt, ideally in a fresh session, that reads the draft as a reviewer: “Check this against our compliance skill and NAR Code of Ethics Article 12, which requires a true picture in advertising. List problems; do not rewrite.” Then a person reads it. For more on making the prose sound less machine-made, see humanizing AI writing.
A similar approach works for any feature-led copy; our product description skills guide covers the claim-check pattern in more detail.
Workflow 2: neighbourhood pages that inform without steering
A good neighbourhood page sticks to verifiable, place-based facts (commute times, parks, transit, housing stock, recent sales data you have the right to publish) and avoids describing the people who live there. That keeps it useful for search and on the right side of 804(c).
Why this is the riskiest page type
Neighbourhood pages invite exactly the language HUD warns about: who lives there, what kind of community it is, whether it is “safe”. A skill should block that and redirect to measurable facts.
Example prompt
Use our neighbourhood-page skill and fair-housing rules.
Area: Riverside district. Data: riverside-facts.md (transit lines,
parks with acreage, drive times to downtown and the airport,
median sale price from our MLS export for the last 12 months,
housing types by count).
Write a 1,200-word page: intro, getting around, parks and
recreation, housing stock, market snapshot with the date range,
FAQ (6 questions), and a CTA to book a tour.
No descriptions of residents. No safety or crime claims.
Cite the data file line for every number.What the output should contain
| Section | Contents | Avoid |
|---|---|---|
| Getting around | Transit lines, drive times with the time of day measured | ”Easy commute” without numbers |
| Parks and amenities | Named parks, acreage, trails, libraries | ”Great for kids” |
| Housing stock | Types, typical lot sizes, eras of construction | ”Starter homes for young couples” |
| Market snapshot | Median price, days on market, date range, data source | Forecasts presented as fact |
| FAQ | Questions buyers search for | Questions about who lives there |
Check your MLS rules before publishing sold data on a public page; boards set their own display rules. For crime and school information, many brokerages link to the official source and let the buyer decide, rather than summarising it. Follow your broker’s policy.
Workflow 3: local SEO for agents and brokerages
Use claude-seo or a similar skill to audit your site, generate RealEstateAgent and LocalBusiness schema, map one page per area you actually serve, and keep your Google Business Profile details consistent with the site. The skill does the checklist work; the local knowledge has to come from you.
What the skill can do well
- Audit title tags, headings and internal links on area pages.
- Generate JSON-LD for
RealEstateAgent,LocalBusiness,BreadcrumbListandFAQPage. Validate it in Google’s Rich Results Test. See schema markup skills. - Plan a page per neighbourhood with a template, which is programmatic SEO on a small scale. Read programmatic SEO with Claude Code before you generate 50 near-identical pages; thin, swapped-name pages are what Google’s spam policies call scaled content abuse.
Example prompt
Use claude-seo. Audit /neighbourhoods/ on our site.
For each page: word count, unique facts vs template text,
title and meta, schema present, internal links in and out.
Flag any page where more than half the text is shared with
another page. Then draft RealEstateAgent JSON-LD using our
office details in office.md.What the output should contain
A table per page, a list of near-duplicate pages, schema that matches your real name, address and phone exactly as they appear on your Google Business Profile, and a short list of fixes ranked by effort. If you pay for keyword data, DataForSEO’s API is pay-as-you-go with a $50 minimum top-up and a $1 trial, which is a cheap way to give a skill real search volumes.
Workflow 4: agent email nurture that respects CAN-SPAM and TCPA
Use an email sequence skill to draft nurture tracks by lead type (buyer, seller, past client, sphere), load your compliance file so every email avoids steering language, and include the CAN-SPAM basics: a real postal address, a clear unsubscribe, and honest subject lines. Texting is a separate set of rules under the TCPA; consent comes first.
Example prompt
Use the email-sequence skill from marketingskills, plus our
fair-housing rules. Lead type: seller who requested a home
valuation. Write 6 emails over 30 days: value report follow-up,
what affects price in their area (from market-data.md), prep
checklist, a recent-sale story using only facts in sold-123.md,
a "questions sellers ask" email, and a soft check-in.
Each email: subject, preview text, 120-180 words, one CTA.
Footer placeholder for our postal address and unsubscribe link.What the output should contain
| Element | Requirement | Source |
|---|---|---|
| Subject line | Not misleading about content | FTC CAN-SPAM compliance guide |
| Identification | Clear that it is from you or your brokerage | FTC CAN-SPAM guide |
| Postal address | A valid physical postal address | FTC CAN-SPAM guide |
| Opt-out | Clear unsubscribe, honoured within 10 business days | FTC CAN-SPAM guide |
| Testimonials | Real, from real clients, with any incentive disclosed | FTC Endorsement Guides, 16 CFR part 255 |
| SMS follow-ups | Prior express consent before automated texts | TCPA; ask your broker or counsel |
Your state licensing rules may also require your brokerage name in every ad, email included. More on building sequences in Claude Code skills for email marketing.
How to build your own real estate compliance skill
Create ~/.claude/skills/fair-housing-rules/SKILL.md, write a clear description so Claude loads it for any listing, neighbourhood, ad or email task, and list your rules with the source for each. It takes about an hour to write and you control every line.
A minimal skeleton
---
name: fair-housing-rules
description: Apply to any real estate listing, neighbourhood page,
ad, social post or email. Enforces Fair Housing Act 804(c),
24 CFR 100.75 and [STATE] protected classes.
---
# Rules
1. Describe the property and place, never the ideal buyer.
2. Protected classes: federal list plus [STATE additions].
3. Neutral property words are fine (HUD 1995 memo).
4. No safety, crime or school-quality claims; link official sources
only if the broker allows.
5. Every fact must trace to the supplied fact file.
6. Output a flags list for anything borderline.
# Sources
42 U.S.C. 3604(c); 24 CFR 100.75; HUD memo 9 Jan 1995.Keep it short. Long rule files get followed less reliably than short, specific ones. Install steps are in how to install Claude Code skills.
Test it with bad inputs
Feed it a fact file that contains “great for families” and “near St. Mary’s church, perfect for parishioners”. A working skill should flag both. If it doesn’t, tighten the description so it loads, and make the rule more explicit.
Common mistakes agents make with AI marketing skills
The most common failures are trusting a skill’s rules without reading them, letting the model invent features, and treating a clean compliance pass as legal clearance. Each one is avoidable.
- Installing a compliance skill unread. As shown above, even a good free skill can contain an unsourced claim and an internal contradiction. Read every line.
- Giving vague inputs. “Nice 3-bed with a big yard” invites invention. Give square footage and lot size.
- One pass for writing and checking. The model that wrote the copy is a weak judge of it. Use a second pass and a human.
- Copying old listing copy as a style sample. If your old listings say “family neighbourhood”, your style sample teaches the problem.
- Publishing MLS data without checking display rules. Boards differ.
- Using a skill built for another country. Australian or UK skills follow different law. US Fair Housing rules apply to US listings.
- Assuming withdrawn guidance is still current, or the reverse. The 2024 digital ad guidance was withdrawn in 2025; the statute was not.
- Mass-producing near-identical area pages. That is a search risk, separate from fair housing.
FAQ
What are the best Claude Code skills for real estate agents?
For most agents: TheProsperityAgent’s free fair-housing-overlay and listing-description, Corey Haines’ marketingskills for email and pages, and claude-seo for local SEO. Add your own state-specific compliance skill.
Can Claude write MLS listing descriptions?
Yes. Give it a verified fact sheet and your MLS character limit, load a compliance skill, and ask for a claim-to-fact table so you can catch invented features.
Is AI-written listing copy a Fair Housing risk?
It can be. The Fair Housing Act applies to the ad, not to who or what wrote it. The agent and brokerage are responsible for what gets published.
What does HUD say about words like “master bedroom” or “family room”?
HUD’s 1995 memo treats facially neutral property descriptions such as “family room” and “master bedroom” as not creating liability on their own. Many brokerages now prefer “primary bedroom” anyway.
Can I mention schools in a listing?
Federal guidance does not ban naming a school as a place. Claims about school quality or who a school suits are where risk rises, and many brokerages restrict school mentions altogether. Follow your broker and MLS.
Is HUD’s 2024 digital advertising guidance still in effect?
No. A Federal Register notice lists it as withdrawn effective 17 September 2025. The Fair Housing Act and 24 CFR 100.75 still apply.
Are there free real estate skills on GitHub?
Yes. TheProsperityAgent/prosperity-real-estate-skills (MIT, 4 skills) and analyticsariel/tech-in-real-estate (MIT, investor-focused) are two. Both are small, so read them first.
Do paid skill packs include real estate skills?
None of the paid packs we verified is real-estate-specific. The Inflectual bundle ($167), Contra’s SEO pack ($39) and SuperSkills ($50) are general marketing packs.
How do I install a real estate skill in Claude Code?
Use /plugin marketplace add then /plugin install for plugin repos, or copy the skill folder into ~/.claude/skills/.
Can Claude Code connect to my MLS?
Not directly in the public skills we found. The connect-mls-to-claude skill works from exports you download. Direct feeds need your MLS’s API access and licence terms.
Can I use Claude to write neighbourhood pages?
Yes, if you give it place-based data and block descriptions of residents and safety claims. Cite a data source and date for every number.
Does Claude Code help with local SEO for realtors?
Yes. Skills like claude-seo audit area pages, generate RealEstateAgent schema and flag near-duplicate pages.
What should an agent nurture email include legally?
Under CAN-SPAM: honest subject lines, clear sender identity, a valid postal address and a working unsubscribe honoured within 10 business days. Your state may add brokerage disclosure rules.
Can I use client reviews in AI-written emails?
Only real ones, with any incentive disclosed, per the FTC Endorsement Guides. Never let the model draft a “testimonial”.
Can I text leads using AI-written scripts?
Automated texts need prior express consent under the TCPA. Get consent first and confirm your setup with your broker or counsel.
How much does it cost to use Claude Code?
Claude Pro is $20 a month, or $17 a month billed yearly, and includes Claude Code. The free skills cost nothing extra.
Did you test these skills on real listings?
No. This guide is based on reading the repos, skill files and official sources. We did not run controlled tests.
Sources and methodology
Repo stars, licences, archived status and last-push dates came from the GitHub API on 2 November 2026. Skill details came from each repo’s README and SKILL.md. Paid prices came from vendor pages or our verified price list; where a price was not shown, we wrote “pricing not public.” We could not fetch Reddit threads for this topic with our tools, so this guide quotes no Reddit users.
- Fair Housing Act section 804(c), 42 U.S.C. 3604(c)
- 24 CFR 100.75 (ecfr.gov). Note: 24 CFR part 109 was removed from the CFR in 1996 (61 FR 14378)
- HUD memorandum, Guidance Regarding Advertisements Under 804(c), 9 January 1995: https://www.hud.gov/sites/documents/DOC_7784.PDF
- Oregon Realtors, permissible advertisements: https://oregonrealtors.org/news/tip-of-the-week/permissible-advertisements-under-the-fair-housing-act/
- Federal Register notice of withdrawal of FHEO guidance documents (effective 17 Sep 2025): https://public-inspection.federalregister.gov/2026-06624.pdf
- FTC, CAN-SPAM Act compliance guide for business; FTC Endorsement Guides, 16 CFR part 255
- NAR Code of Ethics, Article 12
- TheProsperityAgent/prosperity-real-estate-skills: https://github.com/TheProsperityAgent/prosperity-real-estate-skills
- mls-listing-craft: https://claudskills.com/skills/mls-listing-craft/
- coreyhaines31/marketingskills, AgriciDaniel/claude-seo, anthropics/knowledge-work-plugins, anthropics/skills, analyticsariel/tech-in-real-estate (GitHub)
- Inflectual bundle: https://inflectual.gumroad.com/l/master-marketing-claude-skills-bundle
- SuperSkills: https://superskills-one.vercel.app/
- Real Estate Agent Toolkit: https://parclarktech.gumroad.com/l/idtgwm
Your next step
Write your compliance skill first, before any listing skill. Spend an hour on it with the HUD sources above and your state’s protected classes, then test it with two deliberately bad fact files. Once it flags both, install listing-description and run it on one listing you have already published. Compare the two versions with your broker. That single comparison will tell you whether the setup is ready for live listings.
