Resources · Rules API and MCP
Ask your AI assistant about your state’s ad rules
Our library of lawyer advertising rules for every US state and DC is free to use in your own tools. Connect it to Claude, ChatGPT or Cursor and ask about your state’s rules in plain words; the answer comes back with the rule’s source and the date we checked it.
Who is this for?
- Firm owners and marketing staff. Add the connector once and ask your assistant questions like the ones below. It reads our library instead of guessing, and it shows you where each rule comes from.
- Developers and legal-tech builders. Read the same data as plain files, or call the MCP server from your own agent. No key and no account.
- Journalists and researchers. Every rule has a stable link and a one-sentence citation you can quote.
Questions you can ask once it’s connected
- What does a lawyer ad in Texas have to include?
- Which states make lawyers file their ads with the bar?
- Compare the waiting periods for contacting accident victims in Florida, Georgia and Ohio.
- What are New York’s rules on testimonials and past results?
- Has anything changed in the rules this month?
- What have state bars said about paying a lead generator?
- What does Google Ads allow for a bail bond lawyer’s ads?
Use it in your AI tool
The server address is the same everywhere. There is no sign-in: when an app asks how to authenticate, choose none.
https://probativegrowth.com/mcpClaude (the website and the desktop app)
In Claude, open Customize, then Connectors, and add a custom connector. Give it any name, paste the address above, and choose no sign-in. Then ask a question in a new chat. On a Team or Enterprise plan, an owner adds it for the organization first.
Older versions of Claude Desktop that only read a config file can use this in claude_desktop_config.json (it needs Node.js):
{
"mcpServers": {
"legal-marketing-data": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://probativegrowth.com/mcp"
]
}
}
}Claude Code
claude mcp add --transport http legal-marketing-data https://probativegrowth.com/mcpChatGPT
On plans that offer developer mode, turn it on in ChatGPT’s settings, create a connector with the address above, and choose no authentication.
Cursor
Add this to the mcp.json file in Cursor’s settings (or in your project’s .cursor folder):
{
"mcpServers": {
"legal-marketing-data": {
"url": "https://probativegrowth.com/mcp"
}
}
}VS Code with GitHub Copilot
Add this to .vscode/mcp.json in your project:
{
"servers": {
"legal-marketing-data": {
"type": "http",
"url": "https://probativegrowth.com/mcp"
}
}
}Microsoft Copilot Studio
In your agent, add a tool, choose Model Context Protocol, paste the address above, and set authentication to none.
Gemini CLI
Add this to your Gemini CLI settings.json:
{
"mcpServers": {
"legal-marketing-data": {
"httpUrl": "https://probativegrowth.com/mcp"
}
}
}What the assistant can look up
Each tool only reads our data. None of them can change anything, send anything or see your files.
get_state_rules- Every rule we track for one state or DC.
get_rule- One state’s rule on one topic, with the sentence to cite.
compare_states- One topic side by side across the states you name, or all of them.
search_rules- A plain keyword search, such as “Florida filing fee” or “testimonials in New York”. No AI runs on our side.
list_rule_changes- Confirmed changes to the rules, newest first.
data_lookup- One state, platform, metro or court docket in any of our other datasets: bar ad filing, ethics opinions, ad-platform policies, fee caps, federal mass-tort dockets and Spanish-language demand.
data_compare- The same, side by side for up to 10.
data_search- A plain keyword search across those datasets, such as “Google Ads bail bonds” or “lead generator opinion”.
list_datasets- What data the server holds, with the version and DOI of each set.
cite- The citation and credit line for the whole dataset, a state or one rule.
whats_new- What changed recently, across every dataset.
The topics are: ad filing or review, what each ad must include, waiting period for targeted injury contact, results, testimonials and “specialist” claims, referral services and lead generators, ads in Spanish and other languages, call-recording consent, texting and telemarketing law and privacy and health-data law. Every answer carries the rule text, how it was checked (a primary source we read, a secondary source, the ABA model default, or an open question), the sources with the dates we read them, and a link to the rule on its state page.
There is also a ready-made prompt, “Check an ad against a state’s advertising rules”, for apps that show prompts. It loads your state’s rules and asks the assistant to point out which ones your ad touches, as questions for your lawyer, never as a verdict.
The same data as plain files
No server call needed: these are static files, open to any website or script, with no key and no usage cap. Each file starts with the credit line, the citation, the license and the link to the page to cite.
- Every dataset, for people (with how to cite each one)
/data/ - Every dataset
/data/v1/index.json - The rules, by state (Texas shown)
/data/v1/rules/states/tx.json - One topic across every state (filing shown)
/data/v1/rules/topics/filing.json - Confirmed changes
/data/v1/rules/changes.json - Everything, as a spreadsheet file
/data/v1/rules/rules.csv - The credit line and the column guide for that file
/data/v1/rules/README.txt - The machine-readable description (OpenAPI)
/data/v1/openapi.json
The library page has the same data as a table, and each state page has the full text of every rule with its sources.
Credit and license
You can reuse, adapt and republish the data, including commercially, under the Creative Commons Attribution license. The one condition is credit with a link. Paste this line under any chart, table or article that uses it:
Source: Probative, Lawyer advertising rules by state (50 states and DC), version 1.0.0, https://doi.org/10.5281/zenodo.23645751, licensed CC BY 4.0.
To cite a single rule, use the “Cite this” line under that rule on its state page, or ask the assistant to cite it.
The dataset’s permanent record is on Zenodo: https://doi.org/10.5281/zenodo.23645751.
What we keep when you use it
Nothing that identifies you or your firm. We count how many times each tool is used per day, and which kind of app called it, such as Claude or Cursor. We don’t store your questions, the text of your ads, your IP address or any account details. There is no sign-up, so there are no accounts to keep.
To keep the server fast for everyone, very heavy use from one connection is slowed for a minute at a time. The plain files above have no such limit.
Where to go next
Want to see the rules first? Browse the library of lawyer advertising rules by state, or follow changes on Rule Watch. When you’re ready to talk about your own firm’s ads and search, start with the free review below: we handle your state’s ad rules for you.
Where do I start?
See where your firm stands.
Tick what your firm already has: 7 questions, about 2 minutes. You’ll see your stage, what we’d change first, and one place to start, with its price.
Or get a free review: Probative Growth records a 90-second video of what clients see when they look you up, and emails it within 2 business days.