SpreadBench NCSX-IS — AI Infrastructure Bond Issuance Spread Index

AI-infrastructure bond issuance-spread index (NCSX-IS): 13 bonds, 2 tiers, 14 free MCP tools.

使うべきか

品質と安全性

B
説明の品質
98%
スキーマの完全性
57%
命名の品質
80%
ポイズニングのリスク
100%
権限の一致
100%
プロトコルへの準拠
100%

検出事項(1)

  • LOWTool 'ncsx_sector' description lacks action verbncsx_sector 内

ツール定義とプロトコルへの準拠に関する自動分析に基づいています。

コンテキストコスト

~3,295トークン数(ツール定義)
~738 B一般的なレスポンスサイズ
注意への影響は大きい(128k コンテキストの 2.57%)

これは、サーバーのツールがモデルのコンテキストに読み込まれるたびに消費されるおおよそのトークン数です。数が多いほど、ほかのタスクに使える注意が減ります。

インストール

ワンクリックインストール

これを `claude_desktop_config.json` ファイルに追加してください:

{
  "mcpServers": {
    "ncsx": {
      "url": "https://spreadbench.com/mcp"
    }
  }
}

リモートエンドポイント

https://spreadbench.com/mcpstreamable-http

できること

ツール一覧

ツール(14)

🟢 読み取り専用🟡 書き込み🔴 削除⚪ 不明
🟢ncsx_latest

The one number to quote — use this first for "what is NCSX-IS / what is the spread right now". Returns a compact record: headline issuance spread (coupon minus the Treasury yield of matching tenor on the pricing date, weighted by face value), the secured and unsecured sub-indices, constituent/issuer counts, notional, as-of date, methodology version, DOI and data_status. This is a point-in-time issuance spread, not a traded level: there is no daily change. Read data_status before quoting — "real_issuance" means observed public issuance data; anything marked "retired_seed" or "seed" is a demo series and must not be quoted.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_index

The full index definition, not a single number — use when the question is structural ("how is NCSX-IS built", "show all tiers") rather than "what is the level". Returns the NCSX-IS / secured / unsecured tiers side by side with each tier's notional and constituent count, plus index-level coverage notes and data_status. For a quick headline reading use ncsx_latest instead.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_constituents

The whole universe in one response — all 13 constituent bonds, no filter: issuer, ticker, coupon, maturity, size, seniority, pricing date, tenor, the matched Treasury yield on that date, and the resulting issuance spread in bp. Use it to list the index or to compute weights; if the user named one issuer, use ncsx_bond instead so you do not pull 13 rows to answer about one.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_bond(q)

Filtered lookup for one issuer, ticker or note keyword (e.g. "CoreWeave", "GLXY", "Helios", "2031") — returns only the matching constituent bonds, not the whole universe. Constituents are identified by issuer + coupon + maturity; they do NOT carry CUSIPs, so never query by CUSIP. Use ncsx_constituents when you need every bond.

入力スキーマ

{
  "type": "object",
  "properties": {
    "q": {
      "type": "string",
      "description": "Issuer name, ticker or note keyword, case-insensitive substring match"
    }
  },
  "required": [
    "q"
  ],
  "additionalProperties": false
}
🟢ncsx_curve

The US Treasury par-yield curve used on the pricing date (14 tenors, interpolated linearly, never extrapolated) plus its recent history. Every constituent spread is measured against this curve.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_sector

Context series for the AI-infrastructure credit complex: US high-yield OAS, US investment-grade OAS and related ICE BofA indices, with 30-day change and daily history — use these to compare NCSX-IS against the broader market.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_filings(issuer, form)

SEC filing radar for the 9 AI-infrastructure issuers covered by SpreadBench (6 in NCSX-IS plus Nebius, IREN and Hut 8 on the watchlist): every filing from the last 45 days, noise forms (ownership, routine) filtered out, pricing-chain forms (FWP, 424B*, S-3*, F-10) flagged with pricing_signal=true, each row linked to the EDGAR original. NCSX-IS is event-driven, so a new pricing-chain filing is what triggers an index re-version — use this to check whether a re-version is coming.

入力スキーマ

{
  "type": "object",
  "properties": {
    "issuer": {
      "type": "string",
      "description": "Optional filter: issuer name or ticker, case-insensitive substring (e.g. \"CoreWeave\", \"WULF\", \"Hut 8\"). Omit to get all issuers."
    },
    "form": {
      "type": "string",
      "description": "Optional filter: exact SEC form type (e.g. \"8-K\", \"10-Q\", \"424B5\"). Omit to get all forms."
    }
  },
  "additionalProperties": false
}
🟢ncsx_marks

Second, independent layer: quarterly fair-value marks for AI-infrastructure bonds as reported by US registered funds on SEC Form N-PORT (Level 2), converted to spread over the matched Treasury. Compare with NCSX-IS to see issuance spread vs. mark spread.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_wall(year)

Maturity wall for AI-infrastructure bonds: outstanding face value by maturity year, split secured / unsecured, with the full bond list per year. As of 2026-11-04, $17.86B (63.6% of the index) matures in 2031, and 7 of those 8 bonds priced within 2026 — a single issuance window produced a same-maturity wall. Pass year (e.g. 2030) to get just that year: its face value, share of the index, secured/unsecured split and every bond maturing in it. Use for refinancing-concentration and supply-schedule questions. Terms and maturity dates are facts from SEC filings; this is not a refinancing forecast or credit judgement.

入力スキーマ

{
  "type": "object",
  "properties": {
    "year": {
      "type": "string",
      "description": "Optional 4-digit maturity year to drill into, e.g. \"2030\" or \"2031\". Omit for all years plus the peak-year breakdown."
    }
  },
  "additionalProperties": false
}
🟢ncsx_stories

Citable data stories (editorial layer) recomputed live from the same public data as the index, each with headline numbers, bilingual page URLs and a ready-to-post 1200x630 share card: (1) the maturity wall of the peak year; (2) the CoreWeave repricing — every CoreWeave bond widened 191-258bp between issuance and the latest N-PORT fund marks (2026-08-31) while same-sector peers (GLXY / CORZ / CIFR) moved only 17.6-40.7bp. Returns numbers computed from current universe + marks data, not cached text. Facts only — no ratings, no investment advice.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}
🟢ncsx_facts(id, q)

The quotable-facts layer: 16 pre-computed, individually sourced numbers covering the whole index — index and sub-index levels, the secured-vs-unsecured structure premium, universe size, per-bond spread range, peak maturity year and its share, the 2030 window, N-PORT marks coverage and implied delta, the CoreWeave repricing, peer repricing, and NCSX vs US HY OAS with rating-bucket reads. Every fact carries value, unit, as_of, a human-readable calc (so you can restate how it was derived) and a source URL. Use this when the user wants a number to cite, a comparison, or several figures at once — instead of stitching five endpoints together. Filter with id (exact, comma-separated) or q (substring on the label). Pass nothing to get all 16.

入力スキーマ

{
  "type": "object",
  "properties": {
    "id": {
      "type": "string",
      "description": "Exact fact id(s), comma-separated — e.g. \"index_level\" or \"maturity_2030,peak_maturity_year,crwv_repricing\""
    },
    "q": {
      "type": "string",
      "description": "Case-insensitive substring match on fact id and labels — e.g. \"coreweave\", \"2030\", \"maturity\""
    }
  },
  "additionalProperties": false
}
🟢ncsx_events(since, ticker)

The index history — use this when the question is "how did NCSX-IS get to its current level" rather than "what is the level". NCSX-IS is event-driven: it only moves when a new bond prices, so the history is a replay of 13 pricing events from 2025-05-19 to 2026-07-24. Each event returns the date, the index level after the event, the change in bp, notional added, and the per-bond pull attribution (pull_bp = that bond's post-add weight × its spread minus the pre-add index level, so pulls sum to the event delta). Headline: 517.7bp at inception, trough 410.0bp (2026-06-16), 432.3bp now, net -85.4bp over 431 days across 13 events (4 up / 8 down). Filter with since ("2026" or "2026-02-05") or ticker ("CRWV") to see only that window or that issuer's pricing events. There is no daily series and no traded level — do not describe these as closes or weekly moves.

入力スキーマ

{
  "type": "object",
  "properties": {
    "since": {
      "type": "string",
      "description": "Only events on or after this date or year prefix — \"2026\", \"2026-02\" or \"2026-02-05\". Omit for the full history."
    },
    "ticker": {
      "type": "string",
      "description": "Only events that added this issuer's bonds — ticker like \"CRWV\", \"CIFR\", \"APLD\", \"GLXY\", \"WULF\", \"CORZ\". Omit for all issuers."
    }
  },
  "additionalProperties": false
}
🟢ncsx_ruler(coupon_pct, tenor_years, size_mm, seniority, peer_window_years, ...)

Reverse pricing — the question every other tool cannot answer: given a coupon and a tenor, where would a new AI-infrastructure bond land? Returns the matched Treasury yield (linear interpolation between the two adjacent curve tenors, never extrapolated), the resulting issuance spread in bp using the same formula as every constituent ((coupon - UST) x 100), how it compares with the index level and with the same-structure tier, its rank among the 13 constituents, a like-for-like comparable set (same structure, tenor within ±1y, with median / weighted / rank), and — if you pass a size — the index level after adding it (face-value weighted: the same arithmetic as the pull_bp attribution in ncsx_events). Pass issuer to also get that issuer's own bonds already in the index (coupon, tenor, spread at pricing, gap to the hypothetical new bond) — the first thing an underwriter checks is what the same name paid last time. Nothing is stored: it is recomputed on every call from the current curve, universe and index, so it can never go stale. Use it for "is 9.5% for 5 years tight or wide for this market", "what does an 8% coupon imply", "what happens to NCSX-IS if another $2bn prices at this spread", "how would a new CoreWeave bond compare with its existing stack". Facts and arithmetic only: it does not recommend a coupon and issues no rating or investment advice.

入力スキーマ

{
  "type": "object",
  "properties": {
    "coupon_pct": {
      "type": "number",
      "description": "Coupon in percent — 7.5 means 7.5%. Range 0-25."
    },
    "tenor_years": {
      "type": "number",
      "description": "Years to maturity — 5, 5.5, 0.5 (6 months). Must fall inside the published curve range (1M-30Y); out-of-range tenors are refused, never extrapolated."
    },
    "size_mm": {
      "type": "number",
      "description": "Optional: new-issue size in $ millions. Omit or 0 to skip the index-impact calculation."
    },
    "seniority": {
      "type": "string",
      "enum": [
        "secured",
        "unsecured"
      ],
      "description": "Optional: structure of the new bond, default \"secured\". Drives which sub-index the comparison uses (secured 391.2bp / unsecured 540.1bp at the current version)."
    },
    "peer_window_years": {
      "type": "number",
      "description": "Optional: half-width of the tenor window used to build the comparable set (same structure, tenor within ± this many years). Default 1, max 10. Note that all current constituents matured in 5.0-6.1 years, so a tenor far from 5y will fall back to the nearest existing tenor, flagged in peer_set.fallback."
    },
    "issuer": {
      "type": "string",
      "description": "Optional: issuer name or ticker (\"CoreWeave\", \"CRWV\"), case-insensitive substring — same matching rule as ncsx_bond. Returns that issuer's existing index bonds as a pricing history for the new bond to be compared against. Omit to skip."
    }
  },
  "required": [
    "coupon_pct",
    "tenor_years"
  ],
  "additionalProperties": false
}
🟢ncsx_methodology

Methodology metadata and full formula chain: version, DOI, license, data sources and their licensing status, the exact spread and weighting formulas, and what is explicitly NOT measured.

入力スキーマ

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}

コミュニティ

このサーバーを評価する

エビデンス

最近の観測

検証済みバージョンは記録されていませんツール 14 件