Wolfram Alpha MCP server

@pipeworx/wolfram-alpha

Wolfram Alpha MCP — computational, factual, and quantitative queries.

Part of Pipeworx — an MCP gateway connecting AI agents to 1679+ live data sources.

Tools

code in a real kernel (keyless; proxies Wolfram's hosted MCP evaluator).

• short_answer(query, units?) — single terse plain-text answer.
• fullquery(query, units?, includepods?, format?) — structured pods.
• wolframcompute(code, timeconstraint_seconds?) — evaluate Wolfram Language

Auth

• Platform key: gateway env PLATFORMWOLFRAMKEY (shortanswer / fullquery only).
• BYO: ?_apiKey= after registering at https://developer.wolframalpha.com (free 2,000/mo).
• wolfram_compute needs no key.

Evaluation semantics (triage note)

wolframcompute transmits code verbatim — as a JSON field, no URL encoding anywhere in the path — and the kernel evaluates exactly what was sent. A result that disagrees with the intent* of a question is almost certainly a bug in the submitted code, not in transmission or the kernel. Before filing a tool bug, echo the parse back through the tool itself:

wolfram_compute({ code: "ToString[Hold[<the code>], InputForm]" })

Worked incident (fleet #422, 2026-08-17): Length[Select[Permutations[Range[5]], And@@Thread[#!=Range[5]]&]] was reported as "returns 119, should be 44 (derangements of 5)". 119 is the correct value of that code: on two concrete lists, perm != Range[5] evaluates eagerly to a single True/False (whole-list Unequal) before Thread can split it elementwise, so the predicate collapses to "permutation ≠ identity" and counts 120 − 1 = 119. The elementwise form And @@ MapThread[Unequal, {#, Range[5]}] & (or just Subfactorial[5]) returns 44 through the same tool.

Data source

• Short Answers v1: https://api.wolframalpha.com/v1/result
• Full Results v2: https://api.wolframalpha.com/v2/query (output=JSON)
• Wolfram Language kernel: https://agenttools.wolfram.com/mcp (hosted by Wolfram Research)

Quick Start

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

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

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/wolfram-alpha/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — askpipeworx, discovertools, 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 1679+ 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/short_answer \
  -H 'Content-Type: application/json' \
  -d '{"query":"what is the capital of France"}'

Shortened. The full README is on GitHub.