Congress.gov MCP Server

Quick Verdict

Congress.gov MCP Server is an open-source community wrapper maintained by cyanheads. It exposes the Congress.gov API v3 and related official congressional sources through Model Context Protocol. As of July 21, 2026, the current release is v0.4.1, with eleven read-only idempotent tools, five resources, and two research prompts. It is not an MCP product published, reviewed, or endorsed by Congress, the Library of Congress, or Congress.gov. The upstream records are official; the wrapper, package, search mirror, and hosted endpoint are not.

It is useful for journalists, policy researchers, public-affairs teams, and civic developers who need structured discovery. It does not replace authoritative bill text, the Congressional Record, legal advice, or institutional archives. Model summaries can omit amendments, versions, procedural status, and temporal context. Any legal, lobbying, investment, or public-policy decision should return to the original documents and receive qualified human review.

Best For

• Researchers and journalists tracking federal legislation with explicit fields and repeatable filters.
• Civic technology teams connecting bills, votes, members, committees, nominations, and CRS materials to an MCP client.
• Public-affairs teams that can preserve official citations and separate discovery from interpretation.
• Engineers prepared to manage API limits, caching, logs, dependencies, search-index freshness, and regression tests.
• It is not suitable for definitive legal conclusions, guaranteed real-time notice, unreviewed automatic publication, or personal political profiling.

Key Features

• Bills and laws: browse actions, sponsors, summaries, text versions, related legislation, and enacted public or private laws.
• Members, committees, and votes: find members by state or district, retrieve legislative portfolios, inspect committee materials, and read House or Senate roll-call positions.
• Nominations and research records: query presidential nominations, confirmations, CRS bill summaries and policy reports, committee reports, and the daily Congressional Record.
• Optional full-text search: the v0.4 series adds congressgov_search_bills, a local BM25 mirror of bill titles and CRS summaries. It is off by default and requires an explicitly enabled, locally built index.
• Agent context: five resources describe the current Congress, bill types, members, bills, and committees; two prompts structure bill analysis and cross-Congress research.
• Deployment controls: local stdio, Streamable HTTP, and containers, with optional authentication, storage choices, structured logs, and OpenTelemetry.

Use Cases

The server can track recent actions on selected bills, compare public roll-call positions, assemble committee and CRS research, or generate a source list for a reporting lead. The optional mirror addresses a discovery gap in the official API, but it searches bill titles and CRS summaries rather than the entire Congress.gov corpus. Its freshness depends on local synchronization, so it must not be presented as an official or continuously current site search.

Reproducible output should retain the Congress number, bill type and number, text version, query time, tool parameters, and official URL. Similar titles do not imply identical provisions. Introduction, passage in one chamber, passage in both chambers, and enactment are distinct states. CRS material can provide valuable nonpartisan analysis, but it is not a court or legislature’s binding legal interpretation. A model should not infer a member’s motives, ideology, or personal characteristics where the records do not support that conclusion.

Public government records do not make every workflow non-sensitive. Queries can reveal an organization’s policy priorities, an unpublished investigation, or a client’s confidential assignment. Do not place internal notes, personal data, access tokens, or unnecessary client context in a public MCP request. Treat the hosted endpoint, model provider, gateway, and observability stack as separate processors. Store an API key in server-side secret management, never in a prompt or shared workflow.

Pricing

The code uses Apache-2.0, and neither the package nor the community-hosted endpoint carries a software subscription fee. Without a configured key, the server can use DEMO_KEY, documented at 30 requests per hour. A free api.data.gov key is documented at 5,000 requests per hour, but current upstream rules control. Self-hosting adds runtime, mirror storage, index refresh, monitoring, security, and maintenance costs.

Release v0.4.1 fixes the container build for the search-mirror capability introduced in v0.4.0; it does not add a new runtime feature. Pin package or image versions in production, validate provenance, and regression-test upstream field changes. The public endpoint is appropriate for non-sensitive evaluation and does not promise an institutional SLA.

Pros

• Eleven focused read-only tools span bills, laws, members, committees, votes, nominations, CRS, and daily records.
• The opt-in local mirror adds keyword discovery while making its non-official status and setup explicit.
• Resources and research prompts help an agent load legislative context before issuing structured calls.
• DEMO_KEY supports quick validation, while a free official key can support normal research prototypes.
• Apache-2.0 source can be audited and deployed over stdio, HTTP, or containers.
• Stable bill and member identifiers support links back to authoritative records.

Cons

• It is an unofficial community wrapper; government institutions do not guarantee its code, security response, hosting, or conclusions.
• It covers U.S. federal congressional data, not state legislation, case law, administrative rules, or a complete legal research system.
• Full-text search requires local initialization and synchronization, is off by default, and covers only selected text fields.
• Public hosting adds query-log, third-party processing, availability, and supply-chain risk.
• Legislative versions and procedure are complex, and model summaries can flatten legally important differences.
• The demo limit is low, while self-hosted production still requires operational ownership.

Alternatives

FAQ

Is this the official Congress.gov MCP server?

No. It is an unofficial cyanheads community project. It reads official sources, but its code, package, mirror, and public endpoint do not represent Congress or the Library of Congress.

How many tools are in v0.4.1?

There are eleven read-only tools. The bill search tool depends on a local title and CRS-summary mirror, is disabled by default, and needs an initialized index.

Is a CONGRESS_API_KEY mandatory?

No. A low-limit DEMO_KEY is the default. Stable research should use a free key and follow current official rate limits and terms.

Can an MCP summary be treated as legal advice?

No. Verify the bill identifier, text version, action dates, procedural status, and official source. High-risk decisions require qualified professional review.

Does public government data eliminate privacy concerns?

No. Query intent, client assignments, internal annotations, and combined analysis may be confidential even when every retrieved record is public.

Does the local full-text index stay current automatically?

Not by assumption. Operators must build and refresh it, record its update time, and check the official record before publishing a conclusion.

Bottom Line

Congress.gov MCP Server offers a broad structured interface to public federal legislative data, and the v0.4.1 search mirror improves bill discovery for teams willing to operate it. Its durable role is a research and source-location layer, not an official record, legal database, or automated judgment system. Start with low-risk queries, protect API keys, pin and inspect production releases, and document mirror freshness. Every consequential result should preserve the official link, text version, and query date, then receive human review for authority, timing, and legal context.