Dc Code MCP Server

@pipeworx/dc-code

District of Columbia Code (D.C. Code — DC statutes) — full section text by citation, and topic/chapter search over official headings.

Part of Pipeworx — an MCP gateway connecting AI agents to 1784+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Tools

• dc_statute(citation) — full current text of one D.C. Code section (e.g. "22-2101", murder in the first degree), with the official heading, disposition (in force / repealed / transferred / reserved), legislative History annotations, Prior Codifications, and Cross References.
• dc_search(query?, title?, limit?) — topic/keyword search over every section's official heading (e.g. "eviction notice", "concealed carry permit"), optionally scoped to one title. Pass title alone (no query) to list every section under that title — a chapter-list/browse mode. Not full-text search — the DC Council does not publish one.

Auth

Keyless.

Data sources

• https://raw.githubusercontent.com/DCCouncil/law-xml-codified/{branch}/us/dc/council/code/titles/{title}/sections/{citation}.xml — one section's current XML (its heading, body text, and annotations). dc_statute calls this live, per request, pinned to the branch baked below.
• Same repo's us/dc/council/code/index.xml and titles/{title}/index.xml — structure only (title/chapter/subchapter headings + section citation lists, no catchlines). Used only by scripts/bake-index.mjs.
• https://api.github.com/repos/DCCouncil/law-xml-codified/branches — used only by scripts/bake-index.mjs to resolve the current publication branch (see below).

Licence — public domain, and why this pack is allowed to bake an index

code.dccouncil.gov's own footer: "The codes and laws on this website are in the public domain." The same footer: "Please do not scrape. Instead, bulk download the HTML or XML" from the Council's GitHub organization. This pack follows that instruction literally — it never touches code.dccouncil.gov, only raw.githubusercontent.com content from DCCouncil/law-xml-codified. Public-domain content plus the publisher's own explicit redistribution channel means baking a catchline index from this repo needs no separate grant (root CLAUDE.md, "mirror-needs-grant-proxy-does-not" — the licence-permits-it branch, not the silent/forbidden one). Statutory text itself is also never baked — dc_statute always fetches it live, per call.

The default-branch trap

GET /repos/DCCouncil/law-xml-codified reports default_branch: "publication/2021-10-18" — real git metadata, frozen at a 2021 codification, five years stale, even though the repo is actively published: 406 branches existed at bake time, the newest a publication/{base-date}.{amendment-date} pair updated within days. GitHub's "default branch" setting is unrelated to which branch is current. scripts/bake-index.mjs lists every branch (paginated, ~5 unauthenticated requests) and picks the lexicographically-latest name — safe because every name is an ISO-8601 date or a dotted pair of them, so string order is chronological order. The resolved name is baked as DC_CODE_BRANCH and is what dc_statute's live fetch pins to.

Section XML shape

A short section is one flat <text> element; a section with lettered/numbered subsections nests <para> elements, each with its own <num> ("(a)") and <text>, repeatedly — sometimes nested one level deeper. A reason element reading "Repealed" (or Reserved/Transferred) marks a non-substantive stub whose body text is literally "Repealed." — surfaced as disposition rather than left for the caller to infer. A trailing annotations block (History, Prior Codifications, Cross References, Editor's Notes) is excluded from the body-text walk and returned as its own grouped fields.

One parsing trap worth knowing: a num or heading element can carry a NESTED inline element (e.g. Title 7 Chapter 2B's chapter number carries an embedded effective-date annotation tag inside its own num element) — a tight "no-nested-tag" regex silently drops that chapter's number instead of erroring. scripts/bake-index.mjs strips any nested tag content rather than assuming num/heading are always tag-free.

Capabilities — two available, two are not

• Citation lookup — dc_statute. Direct URL construction from the citation; no index lookup needed (title is the citation's own prefix before its first -).
• Topic/chapter search — dc_search, over official headings only (see below). Matches 24,198 sections' real headings, baked once per scripts/bake-index.mjs run — see its header for why per-section fetches were unavoidable (the index tree carries no catchlines, unlike South Dakota's or Oklahoma's siblings).
• Amendments/enactment history — PARTIALLY AVAILABLE. Every amended section's annotations block carries one or more type="History" entries naming the enacting/amending Act; dc_statute returns these as history. There is no single canonical "effective date" field separate from these citations.
• Historical version (the text as it read before an amendment) — NOT AVAILABLE. DCCouncil/law-xml-codified publishes only the CURRENT codified text, no per-amendment archive and no endpoint returning a prior version's text.

Refresh path

DC Council publishes a new amendment branch roughly every few days and a new base codification a few times a year. Re-run node scripts/bake-index.mjs > src/dc-index-data.ts whenever a caller's index_live_mismatch or a stale data_as_of suggests the baked index has fallen behind the live branch — nothing else in the pack needs to change; dc_statute always fetches section text live, pinned to whatever branch the bake most recently resolved. No scheduled workflow exists for this (same as every sibling state-code pack in this family) — it is a manual re-run.

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "dc-code": {
      "url": "https://gateway.pipeworx.io/dc-code/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/dc-code/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1784+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

No MCP client? Call it over HTTP

curl -X POST https://gateway.pipeworx.io/v1/tools/dc_statute \
  -H 'Content-Type: application/json' \
  -d '{"citation":"22-2101"}'

No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/dc_statute. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.

Standalone (no gateway account)

This package also runs as a local stdio MCP server — no Pipeworx account, no gateway round-trip:

{
  "mcpServers": {
    "dc-code": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-dc-code"]
    }
  }
}

Or run it directly to confirm it starts:

npx -y @pipeworx/mcp-dc-code

It speaks MCP over stdin/stdout and answers initialize/tools/list/tools/call for only this pack's tools — none of the shared meta-tools the gateway connection above adds. Same source, same tools, no ask_pipeworx routing.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Dc Code data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides
• pipeworx.io

License

MIT