Ohio Code MCP Server

Ohio Code — Ohio Revised Code by citation

Give it 2903.02, get back the murder statute; or search the whole Revised Code by topic. Keyless.

Part of Pipeworx — an MCP gateway connecting AI agents to 1856+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Caveats

• Fails silently upstream. A nonexistent section returns HTTP 200 with the site chrome and no statute. On the live path the pack keys on the statute container, present on a real section and absent on a miss, so a miss is reported as section_not_found.
• Section numbers are not contiguous; repealed sections are normal (the stored copy lists exactly the sections the Revised Code currently carries).

Tools

oh_revised_code({section: "2903.02"}) → statutory text and effective date. A zero-padded spelling (009.17) and an ORC prefix are read as the same citation as 9.17.

oh_code_search({query: "EDGE certification minority business", limit: 10}) → ranked sections with citation, chapter, heading and an opening excerpt; pass a result's section to oh_revised_code for the full text. It is a full-text match over the stored Revised Code, not semantic search, so use the statute's own words. A query that matches nothing returns found:false; an index that did not answer returns reason: "index_unreachable", which means the infrastructure failed, not that nothing matches.

What text contains (fleet #3598)

text is the section body only, as the Ohio Legislative Service Commission publishes the Revised Code. The effective date comes back separately in effective_date_note. The stored copy is cleaner than the old live read, whose text ran on past the statute into the surrounding page text; nothing from the law itself is missing. Every answer carries a text_scope field: the stored copy (served_from: "stored-copy") is the body alone, while the live fallback returns text from the start of the statute container to the end of the source page and can still carry trailing page text.

Stored copy and data_as_of (fleet #3546)

The Ohio Laws site answers US data centres but stalls or refuses from Singapore: over 7 days external callers saw 10 successes against 61 "down" from there. A cache cannot help a first touch, so the Ohio Revised Code is stored, one JSON object per chapter, in the shared statute store (R2, statutes/oh/) that the Illinois, Indiana, North Carolina, Wisconsin and Wyoming packs also use. The gateway hands the pack its R2 binding and the pack reads only the object a call names (a very large chapter is a small head plus parts, and a call reads the head and one part). A section found there is answered with no upstream request; the response says served_from: "stored-copy" and carries data_as_of (when that chapter was last read from the source). A section the copy cannot answer (chapter missing, object unreadable, or listed with no stored text) falls back to the live read. Absence is never remembered: a chapter published later is served as soon as it exists.

Nothing is loaded at module scope: the data never enters the gateway bundle.

Refresh

node mcps/ohio-code/scripts/bake-chapters.mjs            # all chapters -> mcps/ohio-code/.bake/
node mcps/ohio-code/scripts/bake-chapters.mjs --only X   # one chapter
node mcps/ohio-code/scripts/bake-chapters.mjs --resume   # keep source pages already fetched
node mcps/ohio-code/scripts/upload-chapters.mjs          # publish to R2 (statutes/oh/), changed files only

The source is the Legislative Service Commission's Revised Code; a chapter page carries the text of all of its sections, so the refresh makes one request per chapter (about 1,000), not one per section. Sequential, with a delay between requests, an honest user agent, bounded retries, and a stop on the first 429/403/451. Exit 0 only if every chapter was fetched and parsed and the totals clear the floors (default 900 chapters, 15,000 sections with text); anything less exits 1 with the counts. A chapter whose text is unchanged keeps its previous data_as_of. The baked files are a working copy, not committed (.bake/ is gitignored); upload-chapters.mjs writes only the files whose content hash differs from the one recorded in statutes/oh/_manifest.json, so a refresh that finds nothing new writes nothing. It uses wrangler r2 object put --remote under the operator login; no API token.

Topic search (fleet #3779)

The search index is built from the same baked chapters the section reader serves, so what is searchable is exactly what oh_revised_code returns. It is one full-text index (shared search shard oh-statutes, heading plus text per section). A hit is only an id and a score, so the id encodes the citation and the heading and excerpt are read from the stored chapter; there is no second index to keep in step. Load or refresh it after a bake:

node scripts/ingest-ohio-statutes.mjs      # reads mcps/ohio-code/.bake/, no new crawl

Data source

Ohio Legislative Service Commission, Ohio Revised Code. Ohio statutes are public record.

Quick Start

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

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

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/ohio-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 1856+ 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/oh_revised_code \
  -H 'Content-Type: application/json' \
  -d '{"section":"2903.02"}'

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

Run it from npm

@pipeworx/mcp-ohio-code is a local stdio bridge to the Pipeworx gateway: your MCP client starts it with npx, and it forwards every call to the same https://gateway.pipeworx.io/ohio-code/mcp endpoint shown above. It needs network access; there is no offline mode.

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

Without a key it calls anonymously, capped at 50 calls a day. A free account raises that to 200 a day: get a key at https://pipeworx.io/signup and add "env": { "PIPEWORX_API_KEY": "<your key>" } to the server entry. The key is sent as a Bearer token, the same way a hosted connection sends it, so calls are metered and billed exactly as they are there.

It lists only this pack's tools by default. Set PIPEWORX_META_TOOLS=1 to also list ask_pipeworx and the other gateway-wide tools.

Behind an HTTPS proxy: Node's built-in fetch does not read HTTPS_PROXY by default. Newer Node releases honour it when NODE_USE_ENV_PROXY=1 is set; check your Node version's documentation.

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 Ohio Code data" })

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

More

• Docs and guides
• pipeworx.io

License

MIT