Federal regulations, searchable by AI agents in one query
We built the open-source MCP server that lets AI agents search the Federal Register, eCFR, and regulations.gov in one pass. Now it runs in Gemini for government services, and we build the same thing for agency data.
Ask: Which proposed rules this year would change federal drinking water standards? The agent searches the Federal Register, the eCFR and regulations.gov, then answers with a citation for each source.
Two proposed rules this year would amend 40 CFR Part 141, the national drinking water regulations1. Here is the current text of each affected section2 and the public comments filed on each docket3.
Federal regulatory data is public, and there’s a lot of it
- public comments posted on regulations.gov
- 27M+
- Federal Register pages published in 2024, a record
- 106K
- pages in the Code of Federal Regulations
- 188K+
- CFR titles, updated continuously in the eCFR
- 50
The connector searches all of it in one pass and cites the identifier and link of every record it uses.
Sources: regulations.gov; Office of the Federal Register, 2024 page statistics; Code of Federal Regulations page count, National Archives (2021).
From a plain-English question to the official record
Ask in plain English
An analyst asks the agent a real question, like which proposed rules would change drinking water standards.
The agent writes the query
The server gives the model a small SDK for all three sources. It finds the right calls and runs them in a locked-down sandbox.
Every claim cites its record
The answer links the CFR section, the Federal Register document, and the docket it came from. Anyone can check it.
“This connector lets an agent do that work in one pass and show the official source for every answer. We built it in the open so any agency can read the code and see exactly what it does.”
“By integrating BLEN’s open-source regulatory connector, Gemini for government services helps agencies streamline complex workflows and make informed decisions backed directly by official federal records.”
Open source, so your team can verify it
Every query an agent writes runs in a sandbox with no network, filesystem, or environment access. The only way out is to three government APIs. Your security team can check that for themselves.
- license
- Apache-2.0
- access
- read-only
- sources
- federalregister.gov · ecfr.gov · regulations.gov
- tools
- search_api · describe_schema · execute
- sandbox
- no network, filesystem, or env
Three ways to get started
We also build connectors for SAM.gov contract opportunities, Data.gov, and ClinicalTrials.gov, open source or private to your agency.
- 20+ yearsBuilding digital services and AI tools for federal agencies
- 8(a) · SDVOSB · HUBZoneSBA-certified small business
- GSA MASMultiple Award Schedule contract holder
- Washington, DCA team that works where you do
Questions agencies ask us
What is an MCP server?
The Model Context Protocol (MCP) is an open standard that lets AI assistants use outside tools and data. An MCP server is the piece that exposes one data source, like the Federal Register, to any compatible agent.
Can the connector change federal records?
No. It is read-only. Agent code runs in a sandbox with no network, filesystem, or environment access, and can only reach the Federal Register, eCFR, and regulations.gov APIs.
Do we have to use Gemini for government services?
It’s available in Gemini for government services today. Because it follows the open MCP standard, it also works with other MCP clients, and you can self-host it.
Is it really open source?
Yes. It’s released under the Apache-2.0 license at github.com/blen-labs/fedreg-mcp-server, including the security threat model.
Can BLEN build an MCP server for our own systems?
Yes. We build connectors for public federal data and for agency systems, open source or private. Start with a scoping call and we’ll tell you what’s feasible.
How can our agency work with BLEN?
BLEN holds a GSA Multiple Award Schedule contract and is an SBA-certified 8(a), Service-Disabled Veteran-Owned, and HUBZone small business.
Tell us what you need to answer
A regulatory question, or data your agents can’t reach yet. The engineers who built this connector run the call.
Washington, DC · 20+ years in federal