Connect ChatGPT to federal regulatory data via MCP

ChatGPT can't browse regulations.gov on its own, and the raw data would be hard to use even if it could. A single high-profile rule can draw tens of thousands of public comments, most of them mass form-letter campaigns, with the substantive arguments buried inside attached PDFs. There is too much of it to paste into a chat window.

Government Context puts that data behind a single MCP server. MCP is an open standard, not an Anthropic product, so the same server that connects to Claude also connects to ChatGPT. It covers federal dockets, proposed and final rules, and public comments that have been analyzed for stance and commenter type — plus the wider record around a rule: bills and votes in Congress, committee hearings, agency press releases, grant opportunities, lobbying disclosures, and federal awards. Once ChatGPT is connected, you can ask it something like "which trade associations oppose the FMCSA broker-transparency rule, and what are they arguing?" and get an answer grounded in the actual comment record.

This guide shows you how to connect it. Setup takes a few minutes, and the free tier is enough to follow every step. You don't need a credit card.

What you'll be able to do

Once ChatGPT is connected, it has a set of twenty tools (all prefixed gc_) for working with the federal policy record. You never call these by name. You ask in plain language and ChatGPT picks the right tool:

The full tool set goes further. gc_list_comments filters a rule's comments by stance, commenter type, date, or attachments, and gc_get_comment reads one in full. gc_list_activity reports what has happened recently — documents posted, comment deadlines, press releases, grant postings, bill actions — for an agency, topic, bill, or anything else you track. A family of lookup tools (gc_get_bill, gc_get_legislator, gc_get_hearing, gc_get_agency, gc_get_press_release, and the rest) pulls the details of any record gc_search finds, and gc_list_follows and gc_unfollow manage your alert subscriptions.

What MCP is

MCP (the Model Context Protocol) is an open standard for connecting AI assistants to outside tools and data. Government Context runs an MCP server; ChatGPT is one of several MCP clients that can connect to it. Connecting the two is a one-time setup. Because MCP is a standard, the same server works with Claude and Cursor too. If you use Claude, see how to give Claude access to federal rulemaking data.

Before you start

Two things:

A ChatGPT plan that supports custom connectors. Adding your own MCP server is a paid-plan feature. It's available on ChatGPT Plus, Pro, Business, Enterprise, and Edu, not on the free ChatGPT tier. If you don't see the connector settings described below, that's usually why. (This is a limit on ChatGPT's side, not Government Context's; the Government Context account itself is free.)

A free Government Context account. The MCP server requires it so your connection is authenticated, and it's how any alerts you set up find their way back to you. Signup is free:

1. Go to governmentcontext.com and create an account.
2. New accounts start on the free tier. You don't need a card.

You'll sign in through this same account when you connect ChatGPT below. The connection uses OAuth, so you never copy an API key or paste a token anywhere.

The free tier covers search, rule details, comment reading, basic filters, and the corpus overview, on data up to one release cycle behind (public comments become visible after seven days). The paid plan adds the freshest data and the deeper analytics, including campaign and issue drill-downs and the position map. See free vs. paid below.

Connect ChatGPT

The MCP server URL is the same everywhere:

https://mcp.governmentcontext.com

ChatGPT reaches an MCP server's full set of tools through developer mode, which exposes custom connectors' read and write actions in a normal chat. Turn it on, then add the connector:

1. Open Settings, then Connectors.
2. Under Advanced, enable Developer mode. (On Business, Enterprise, and Edu workspaces, an admin may need to allow custom connectors first.)
3. Back on the Connectors screen, click Create (or Add custom connector).
4. Give it a name like Government Context and paste the server URL: https://mcp.governmentcontext.com
5. Save. ChatGPT opens a browser step to sign in to your Government Context account and approve access. This is the one-time OAuth handshake, with no token to copy.
6. Start a new chat. Open the tools menu (the + in the message box), and enable the Government Context connector for that conversation.

ChatGPT can now call the gc_ tools whenever a question needs them.

A note on ChatGPT's "connectors" for deep research. Outside developer mode, ChatGPT's connectors are limited to search-and-fetch for its research feature, which surfaces documents but won't run the analytical tools like gc_analyze_comments or set up a follow. Developer mode is what unlocks the full tool set described here. If a connector seems to only find and quote documents but never analyze a comment corpus, check that developer mode is on and the connector is enabled for the chat.

Your first query

With the connector enabled for the chat, try a query like this:

Which trade associations oppose the FMCSA broker-transparency rule, and what are their main arguments?

ChatGPT searches for the rule (gc_search), pulls its details and comment overview (gc_get_rulemaking), filters the comments to trade-association commenters that oppose it (gc_list_comments), and reads a few in full (gc_get_comment) before summarizing the arguments and naming the commenters it drew from. On regulations.gov the same task means reading comments by hand for an afternoon.

If a result is limited by the free tier, ChatGPT will say so. The server returns a short note explaining what was withheld and that the paid plan includes it, so ChatGPT can relay that instead of guessing.

More things to ask

A few prompts that map cleanly onto what's connected:

• "Give me a breakdown of the comments on Docket EPA-HQ-OAR-2023-0234. How many support versus oppose, and by what kind of commenter?" ChatGPT runs the corpus analysis (gc_analyze_comments) and narrates the structure it returns.

• "Are the comments on this rule mostly a form-letter campaign or individually written?" This uses the campaign-versus-organic detection, so mass campaigns don't drown out the substantive comments.

• "What's the comment deadline on this proposed rule, and what stage is it at?" Comment-period status and regulatory stage come from gc_get_rulemaking.

• "What happened on rules from the Department of Transportation between two and one weeks ago?" This asks for a fully visible free-tier window in the recent-activity feed for an agency (gc_list_activity), which also covers press releases, grant postings, and bill actions.

• "Is anything moving in Congress on broker transparency? Any bills or hearings?" Legislative records come from the same search plus the lookup tools (gc_get_bill, gc_get_hearing).

• "Follow this docket and alert me weekly when new comments are posted." This creates a subscription (gc_follow) tied to your account, delivered as an email digest. Follows can target a rulemaking, a topic, an agency, an organization, a bill, a legislator, or a committee.

Free vs. paid

Free is one release cycle behind.

The free tier gives you search, rule and bill details, comment reading, and basic filters (stance, commenter type, date), plus the corpus overview, up to 25 active follows with daily or weekly alerts, and 100 tool calls per hour. Continuously updated sources — public comments, press releases, legislative and award events — become visible after a seven-day delay, and quarterly disclosures (lobbying, political giving) show the previous complete reporting period.

The paid plan ($500/mo per organization, currently labeled "Beta pricing") includes everything the free tier sees, plus current data with no delay, the latest lobbying and political-giving periods, federal grant opportunities, the deeper analytics (campaign and issue drill-downs, the position map), instant alerts, team seats, and 1,000 tool calls per hour.

Most gated list and overview calls return a useful visible subset and state what a paid plan would add. Paid-only filters (campaigns and issues), grants, and instant alert cadence return explicit upgrade guidance instead of silently changing the request. You can start free and upgrade when your team needs them.

To connect, add the server at https://mcp.governmentcontext.com and sign in with a free Government Context account. You don't need a card.