regulatory-catalysts

Keyless biotech regulatory catalysts: FDA advisory-committee (AdCom) meetings, the evidence those panels review, and disclosed PDUFA decision dates.

Part of Pipeworx — an MCP gateway connecting AI agents to 1764+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

fda_adcom_materials (fleet #2793)

Pick a meeting by committee (acronym or name fragment: ODAC, VRBPAC, CTGTAC, oncologic, gene therap…), date (2026-04-30, 2026-04 or 2026), and/or query (drug, sponsor, application or topic). With no filters it returns the meeting nearest to today. limit (1–3) returns more than one meeting; include_questions_text: false skips the questions-PDF fetch.

How a query is resolved, in order, with matched_by saying which won:

1. title — the calendar's meeting titles (they name the committee, rarely a drug).

2. federal_register — the Federal Register notice that announced the meeting names the application and the drug; one full-text search plus up to three notice bodies yields the meeting date(s), which select the calendar row.

3. agenda — the 20 meetings nearest to today are opened and their page text searched. Last resort, bounded.

Each meeting carries: committee, center, meeting_date, time, the page url, page_updated, the public docket (+ regulations.gov URL), webcast_url, the agenda text, and applications parsed from the agenda (NDA 220359, product, sponsor, proposed indication). Then either:

• materials_status: "posted" with materials_count, materials_by_type, the materials list (title, document_type, party, file, media_id, url), and voting_questions — the Questions PDF's text and its numbered items with a vote flag — when FDA has posted a Questions document; or

• materials_status: "none_posted" with materials_count: 0 and a materials_note carrying the meeting date and FDA's posting timeline. There is deliberately no materials: [] here: nothing posted yet is a statement about the meeting, not an empty search result.

document_type is one of briefing_document, agenda, questions, roster, presentations, minutes, transcript, webcast_information, federal_register_notice, summary, public_comments, other. party is fda / sponsor / combined only when the title says so, otherwise null — CDC, academic and patient-group decks sit on the same pages and a guessed "sponsor" would read as a fact.

What FDA does not publish, so neither do we

• Per-document posting dates. The page's own "content current as of" stamp is returned as materials_as_of; it is the closest honest answer.

• The vote tally. It is in the minutes and the transcript, which are linked, not in any structured field on the page.

Source and failure policy

• Calendar: https://www.fda.gov/datatables-json/advisory-committee-calendar-json — the JSON route behind the calendar page's DataTable (found in the page's own JS bundle; the page itself renders no rows server-side). One row per meeting, 2016 → the furthest scheduled meeting, ~130 KB, cached at the edge for an hour.

• Meeting pages and /media/<id>/download PDFs on www.fda.gov.

• Edge-probed 2026-10-07 with a throwaway wrangler dev --remote Worker on the prod account: the calendar JSON, two meeting pages and a 126 KB questions PDF all answered 200 with this pack's UA. The Orange Book refusal recorded in mcps/fda-drug-competition is specific to the bulk asset route; the small per-meeting PDFs here served from the edge.

• A non-2xx from any of these, a calendar body that is not a non-empty JSON list, a meeting URL that serves fda.gov's "FDA Internet Site Error" apology page, or an Event Materials table whose rows no longer parse is a loud { error } naming the upstream status and the page — never found: false, never an empty list. The questions-PDF step is the one best-effort part: its failure is reported in-band as voting_questions.extraction_error, beside the document's URL, so the rest of the answer stands. src/index.test.ts pins each of these.

• found: false is reserved for filters that genuinely match no meeting, and carries the calendar's size and date range so a caller can tell "not held" from "not in the window".

Smoke, 2026-10-07 (laptop, esbuild bundle of this file): {committee:"ODAC"} → April 30, 2026 ODAC, 16 materials (4 briefing documents, 4 presentation decks, 2 transcripts, minutes, agenda, questions, 2 rosters, webcast), 2 applications (NDA 220359 camizestrant, sNDA 218197/S-004 capivasertib, both AstraZeneca), 2 VOTE questions extracted; {query:"camizestrant"} → the same meeting via federal_register; {committee:"microbiology devices", date:"2026-11-19"} → none_posted with the date; {committee:"VRBPAC"} → October 1, 2026, 9 materials, the strain-selection voting question.

Related MCP server: mcp-kalshi

Quick Start

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

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

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/regulatory-catalysts/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 1764+ 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/fda_adcom_calendar \
  -H 'Content-Type: application/json' \
  -d '{}'

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

Or run it directly to confirm it starts:

npx -y @pipeworx/mcp-regulatory-catalysts

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 Regulatory Catalysts data" })

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

More

• Docs and guides

• pipeworx.io

License

MIT