Pump the federal record straight into your agent.
Your agent is only as good as its context. Connect once over MCP and it can search, analyze, and monitor the same continuously synced government record you read — instead of guessing from its training data.
claude mcp add --transport http government-context https://mcp.governmentcontext.comWorks with Claude Code, claude.ai, and Slack — connection guides
Work your agent can actually do.
Twenty tools cover search, activity timelines, comment analytics, full records, and follows. Here's what one sitting looks like:
- 1
Research a rulemaking
Point your agent at a docket and get the shape of the fight in minutes — who filed, and what the record around the rule looks like.
- 2
Brief before the meeting
Ask for the breakdown and get positions, commenter types, and campaigns as structured data — with links to every underlying filing so the brief survives scrutiny.
- 3
Monitor as it happens
Have your agent set up follows on what you cover. New developments arrive in your feed and inbox, and your agent can pick any of them up and dig in.
> What's the OCC capital rule docket, and how active is it?
⏺ gc_get_rulemaking(rulemaking_id: "OCC-2026-0134-0001")
⏺ Proposed rule, comment period open. 1,847 comments so far — 612 businesses, 41 trade associations, 9 form-letter campaigns.
> Break down the business opposition for my briefing.
⏺ gc_analyze_comments(rulemaking_id: "OCC-2026-0134-0001", commenter_types: ["business"], posted_before: "2026-07-25")
⏺ Of 583 business comments in the visible window, 71% oppose. The opposition centers on the risk-weight floor: · Harbor Point Bancorp — oppose — "the floor would raise costs on…" · Meridian Freight Systems — oppose — "treats trade receivables as…" One campaign: 214 near-identical letters via a trade association.
> Keep an eye on this for me going forward.
⏺ gc_follow(target: "topic:bank-capital", cadence: "daily")
⏺ Following bank capital, daily digest. New comments, hearings, and filings on this topic will reach your feed and inbox.
Wherever your agent lives.
Claude Code
One command in your terminal. Best for deep research sessions and building on the data.
claude.ai
The same connector, chat-shaped — ask questions in plain language from any browser.
Slack
Claude in the channel where your team already works, answering with the same record.
The free tier is the product, delayed.
A free account gets your agent 100 tool calls an hour against the real record, one release cycle behind — continuously updated data appears after a seven-day delay, and some deep analysis is paid-only. Paid removes the delay, adds grants and full comment analytics, and raises the limit to 1,000 calls an hour.
Compare free and paid →Connected safely
Connecting uses standard OAuth sign-in — no API keys to paste into config files. Tools act only within your account's own permissions, and you can revoke an agent's access at any time. Every answer stays checkable: your agent can always link you to the page and the original government record behind what it found.
How authentication works →Give your agent the context it needs.
Create a free account, run the install command, and ask your first question.