Usgs Mineral Commodities MCP Server

@pipeworx/usgs-mineral-commodities

USGS Mineral Commodity Summaries (MCS) — U.S. production, imports, exports, consumption, price, employment and net import reliance for 85 nonfuel mineral commodities (lithium, copper, cobalt, rare earths, gold, aluminum, nickel, platinum-group metals, and dozens of industrial minerals), plus world mine/plant production and reserves by country for 75 of them.

Part of Pipeworx — an MCP gateway connecting AI agents to 1721+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Tools

• mcs_list_commodities(query?, limit?) — browse the 85 commodities this release covers; each entry names its own stat columns (every commodity's MCS chapter defines its own table) and whether a world comparison table (mcs_country_comparison) is available for it.
• mcs_commodity_data(commodity) — one commodity's U.S. salient statistics: the most recent 5 years of production, imports, exports, consumption, price, employment, and NET IMPORT RELIANCE — the share of U.S. consumption that comes from imports. Accepts a common name ("lithium", "rare earths", "platinum group metals") or the MCS code.
• mcs_country_comparison(commodity, countries?, limit?) — world mine/plant production (prior-year actual + current-year estimate) and reserves, by country, for one commodity — "which countries hold lithium reserves", "China vs. Australia rare-earth production". Covers 75 of 85 commodities (see "Known gaps" below for the other 10).

Auth

Keyless.

Data sources

USGS's National Minerals Information Center publishes MCS once a year as a file-only ScienceBase data release — verified live 2026-10-07: ScienceBase item 677eaf95d34e760b392c4970 ("U.S. Geological Survey Mineral Commodity Summaries 2025 Data Release, ver. 2.0, April 2025") carries two ZIPs of CSV tables and a revision-notes .txt — no JSON/REST endpoint of any kind:

• Salient_Commodity_Data_Release_Grouped_MCS_2025.zip — one mcs2025-<code>_salient.csv per commodity (85 commodities), U.S. salient statistics for the most recent 5 years. Column names vary per commodity.
• World_Data_Release_MCS_2025.zip — one MCS2025_World_Data.csv, world mine/plant production (2023 actual + 2024 estimate in this release) and 2024 reserves, by country, for ~79 commodity groupings (coarser than the 85 salient codes).

scripts/bake-index.mjs downloads both, parses every CSV, matches salient commodities to world-table rows by name (with a short hand-checked alias list for naming differences between the two publications — see the script's own comments, including one USGS typo: "Germanium" vs. the world table's "Gemanium"), and writes src/mcs-index-data.ts.

Why this bakes a table in rather than calling a live endpoint

The MCS data release is file-only — there is no API to proxy. Root CLAUDE.md's standing guidance covers exactly this shape (a source that only publishes bulk files), and US federal data (USGS) is public domain, so baking the release's own published numbers into this pack needs no reuse grant. Refresh path: USGS publishes a new MCS release every January (sometimes revised again in spring, as 2025's "ver. 2.0" was) — re-run node scripts/bake-index.mjs > src/mcs-index-data.ts against the new release's ScienceBase item id when one is published.

Why the baked table isn't a static import

The table is ~430 salient rows + 1,250 world rows (~509KB as JSON). A static import would put the whole thing into every gateway isolate's heap whether or not that isolate ever serves this pack — exactly the failure six state-code packs caused on 2026-10-07 (fleet #2754: ~48MB of retained heap and about one gateway call in four dying as a Cloudflare 1102). Instead the table is uploaded to KV at deploy (workers/gateway/src/pack-baked-indexes.json registers it; scripts/sync-gateway-static-json.mjs uploads it) and the gateway injects the parsed object as args._bakedIndex on the first call into this pack — adopted once per isolate. A missing or malformed injection throws loudly (see src/index.test.ts) rather than answering as if a commodity or country were simply not found.

Known gaps

• 10 of the 85 commodities have no row in the MCS world production/reserves table in this release — mostly scrap/slag/crushed-stone categories USGS tracks only domestically, plus a few rare elements (scandium, thallium, thorium) with no global reserve estimate published: Iron and Steel Scrap, Iron and Steel Slag, Quartz Crystal (Industrial), Sand and Gravel (Construction), Scandium, Stone (Crushed), Stone (Dimension), Thallium, Thorium, Yttrium. mcs_country_comparison returns found: false, reason: "no_world_data" for these rather than an empty row list.
• Raw USGS footnote codes (W = withheld to avoid disclosing company proprietary data, E = estimated, NA = not available, ">50" on net import reliance, free-text notes like "Less than 1/2 unit.") are passed through verbatim, never coerced to numbers or nulled out — every tool's response includes a footnote_legend field.
• This is a point-in-time annual release, not a live feed. Every response carries data_as_of (the bake's capture timestamp) and release_year.

Quick Start

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

{
  "mcpServers": {
    "usgs-mineral-commodities": {
      "url": "https://gateway.pipeworx.io/usgs-mineral-commodities/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/usgs-mineral-commodities/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 1721+ 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/mcs_list_commodities \
  -H 'Content-Type: application/json' \
  -d '{"query":"lithium"}'

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

Or run it directly to confirm it starts:

npx -y @pipeworx/mcp-usgs-mineral-commodities

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 Usgs Mineral Commodities data" })

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

More

• Docs and guides
• pipeworx.io

License

MIT