Ofr Funding MCP Server

@pipeworx/ofr-funding

The Office of Financial Research (US Treasury) Short-Term Funding Monitor API — collateral- and tenor-level US repo market rates/volumes (DVP, GCF, tri-party) and money-market-fund balance-sheet composition, neither of which FRED or the NY Fed Desk's own release carries.

Part of Pipeworx — an MCP gateway connecting AI agents to 1867+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Tools

• ofr_repo_rates(collateral, tenor, measure, vintage, start_date, end_date, limit) — repo rates or volumes for one collateral segment (dvp/gcf/tri/tri_ex_fed) x tenor (overnight/term bucket/collateral-type split) x measure (rate/volume/outstanding).
• ofr_mmf_holdings(category, start_date, end_date, limit) — monthly money-market-fund investment composition (repo by collateral/counterparty type, outright Treasuries/agencies, bank-related assets).
• ofr_search_funding_series(query, dataset, limit) — name/mnemonic lookup across the full 206-series repo+mmf catalog (OFR has no live search endpoint).
• ofr_funding_anomaly(mnemonic OR collateral/tenor/measure/vintage, window_days) — compares the latest observation to a trailing window and reports delta + z-score, stating the exact window used.

Auth

Keyless. No registration, no API key, no rate-limit headers observed.

Data sources

• https://data.financialresearch.gov/v1/series/timeseries?mnemonic=... — the only live per-query route. Returns a bare [[date, value], ...] array for one mnemonic. Used by all four tools for every actual data point returned.
• https://data.financialresearch.gov/v1/series/dataset?dataset=repo and ?dataset=mmf — bulk metadata+data dump (~9MB gzip'd, ~450KB for mmf) used ONCE (2026-09-23) to generate src/catalog.ts (mnemonic → human name → start date → frequency). Never called at request time; too large to fetch per call in a Worker, and would be a bulk mirror if it were. The actual VALUES this pack returns always come from the timeseries endpoint above, live, per request.
• https://www.financialresearch.gov/short-term-funding-monitor/ — the human-facing site these APIs back.

Traps

• /v1/series/search, /v1/series/metadata, /v1/metadata/series, /v1/series/datasets all answer 403 {"message":"Missing Authentication Token"}. This is API Gateway's stock response for an undefined route, not an actual auth requirement — there is no public discovery/search endpoint at all. That's why ofr_search_funding_series exists as a static, generated lookup table instead of a live search call.
• Mnemonic grammar: REPO-{COLLATERAL}_{MEASURE}_{TENOR}-{VINTAGE} for repo, MMF-{...}-M for MMF. Not every collateral/tenor/measure combination exists — e.g. DVP has no Treasury/Agency/Corporate collateral-type split, only GCF/TRI do. An invalid combination 400s upstream with body Invalid mnemonic; ofr_repo_rates surfaces that with a pointer to ofr_search_funding_series rather than guessing validity client-side.
• Vintage matters: OFR publishes a preliminary value first and a final revision a few days later, as a SEPARATE mnemonic suffix (-P vs -F), not an in-place update. Defaulting to preliminary gets you the freshest number; final is more accurate for anything more than a few days old.
• Reporting lag: the latest repo-rate observation is typically 1-2 calendar days behind today (e.g. probed 2026-09-23, latest tri-party overnight rate was dated 2026-09-21) — this is normal publication lag, not an outage.
• ofr_funding_anomaly's z-score is computed by this pack, not an OFR-published statistic — it excludes the latest point from its own trailing window so the window and the point being judged never overlap.

Quick Start

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

{
  "mcpServers": {
    "ofr-funding": {
      "url": "https://gateway.pipeworx.io/ofr-funding/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/ofr-funding/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 1867+ 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/ofr_repo_rates \
  -H 'Content-Type: application/json' \
  -d '{"collateral":"tri","tenor":"overnight","measure":"rate"}'

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

Run it from npm

@pipeworx/mcp-ofr-funding 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/ofr-funding/mcp endpoint shown above. It needs network access; there is no offline mode.

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

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 Ofr Funding data" })

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

More

• Docs and guides
• pipeworx.io

License

MIT