mcp-delaware-code

@pipeworx/delaware-code

Delaware Code — state statutes by citation, full-text search, and amendment history. Keyless.

Part of Pipeworx — an MCP gateway connecting AI agents to 1855+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Tools

• de_statute(citation) — full text of a Delaware Code section by citation ("11-636" is murder in the first degree; "6-2734" is contracts with automatic renewal provisions). Resolves the citation to its one chapter/subchapter page via a baked index, fetches that page live, and slices out the section's text plus its amendment history.

• de_search(query, limit?) — real full-text search over Delaware Code statutory text, via the Delaware General Assembly's own live search API (not a baked catchline index — see below).

Auth

Keyless.

The survey trap was half right

The recorded trap for Delaware was "indexes name chapters but publish no section ranges" — true: a citation like "11 Del. C. 636" does not say which of Delaware's ~2,200 chapter/subchapter HTML pages holds it, and unlike North Dakota there is no single combined index page to bake from. But every one of those ~2,200 pages is small, plain, keyless HTML (no PDF, no login, no JS rendering required), and the whole tree is walkable:

https://delcode.delaware.gov/index.html                     -> 31 titles
https://delcode.delaware.gov/title11/index.html               -> chapters
https://delcode.delaware.gov/title11/c005/index.html           -> either
    section content directly, or a list of subchapter links
https://delcode.delaware.gov/title11/c005/sc02/index.html    -> section
    content, when the chapter has subchapters

So the fix is the same shape as mcps/north-dakota-code (crawl once, bake a citation index, fetch live per call) even though the underlying page layout is completely different — see scripts/bake-index.mjs for the crawl and src/de-index-data.ts (generated) for the baked table. The table stores citation + official catchline + chapter/subchapter/title context + the ONE content-page URL for that citation — never the statutory text itself, which de_statute fetches live on every call.

Refresh path

Re-run the bake script whenever Delaware's General Assembly recodifies (adds, renumbers, or splits a chapter/subchapter):

node mcps/delaware-code/scripts/bake-index.mjs > mcps/delaware-code/src/de-index-data.ts

It fails loudly if the root index parses to fewer than 25 titles, or the whole crawl yields suspiciously few content pages or sections — both would mean delcode.delaware.gov changed its markup rather than that the Code got smaller. It also prints one WARNING: line per chapter/subchapter it found with no section content (reserved placeholder chapters, or a chapter transferred elsewhere whose old page carries nothing) — these are skipped, not treated as failures, because the on-site index itself still lists them.

de_search calls a REAL full-text search API, not a baked catchline index

Unlike the ND/IA/KY sibling packs, Delaware's Legislature runs a genuine full-text search behind its on-site search box — found only by reading the search page's own Kendo grid configuration (there is no documented public endpoint):

POST https://legis.delaware.gov/json/DelCodeSearch/GetDelCodeSearchResults?searchTerm=<q>

Verified live: calling it cold, with no prior page load, no cookie, and no anti-forgery token, still returns results — it is stateless. A search for "landlord" returns 308 total matches across the whole Code, including sections whose own title never contains the word "landlord" (it matches statutory TEXT, not just section captions). de_search therefore calls this endpoint directly instead of falling back to a baked catchline index — the baked table in this pack exists only to resolve a citation to its one content page for de_statute, not to support search.

Slicing a section and its history

A leaf page's markup wraps each section as:

<div class="Section">
  <div class="SectionHead" id="636">§ 636. Murder in the first degree; class A felony.</div>
  <p class="subsection">(a) A person is guilty of murder in the first degree when:</p>
  ...
  11 Del. C. 1953, § 636; <a href="...">58 Del. Laws, c. 497, § 1</a>; ...
</div><br><div class="Section">
  ... next section ...

The statutory text is every between one `SectionHead` and the next. The **amendment history** is the plain-text + citation chain that follows the last , up to the that closes the Section wrapper — that closing pair is the one reliable boundary, verified to hold both for a middle section (follows) and the page's last section ( follows, into the page's own closing tags). Reserved placeholder sections (e.g. "637-640. [Reserved.]") have no `` at all, so text and history both come back empty — correctly, not as a parse failure.

Four capabilities — three available, one is not

• Section by citation — available (de_statute).

• Topic/keyword search — available (de_search), and it is REAL full-text search (see above), not caption-only.

• Amendments/history — available, inside de_statute's response (history field): the session-law citation chain Delaware prints at the end of every section, whether or not it is still in force.

• Historical version (a section's text as it read before a specific amendment) — not available. delcode.delaware.gov publishes only the current online edition; there is no per-year archive the way Iowa or Indiana publish. A small number of sections carry a forward-looking [Effective until ] / [Effective ] label pair when a future amendment is already scheduled (both current and prospective text are then both "current," just not yet both in force) — that is the live page's own labeling, not a historical archive, and de_search passes those labels through verbatim when present.

Citations

Delaware cites by title-section, e.g. "11-636" (Title 11, Section 636: murder in the first degree) or "6-2734" (Title 6, Section 2734: automatic-renewal contracts). de_statute also accepts "11 Del. C. § 636", "11 Del. C. 636", and "Title 11 § 636". A small number of sections carry a letter suffix ("11-607A") or are indexed under a shared id for a reserved range ("637-640") — de_statute expands ranges and lets a letter-suffixed citation be looked up directly against the baked index, matching however delcode.delaware.gov itself prints the section id.

Data sources

• https://delcode.delaware.gov/index.html and its title/chapter/ subchapter index pages — crawled once into src/de-index-data.ts (citation, catchline, chapter/subchapter/title context, and the one content-page URL per citation); never fetched per call.

• https://delcode.delaware.gov/title/.../index.html — the one chapter or subchapter page holding a given section's full text. de_statute fetches this live on every call.

• https://legis.delaware.gov/json/DelCodeSearch/GetDelCodeSearchResults — the Delaware General Assembly's own full-text search API. de_search calls this live on every call.

All three are plain, keyless, public — no login, no API key, no rate limit observed during development.

Quick Start

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

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

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/delaware-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 1855+ 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/de_statute \
  -H 'Content-Type: application/json' \
  -d '{"citation":"11-636"}'

No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/de_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": {
    "delaware-code": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-delaware-code"]
    }
  }
}

Or run it directly to confirm it starts:

npx -y @pipeworx/mcp-delaware-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