Seiche

MCP金融服务

提供方 beepboop2025

金融服务 未标注 OAuth 授权 社区 / 第三方 ClaudeCursorGemini CLICodex
查看接入方式

Seiche: free open source funding stress terminal for US money markets. 22 engines, honest backtests, zero data cost (Fed, NY Fed, OFR, Treasury public APIs). AGPL-3.0.

一句话选型结论

按资料字段整理

Seiche 属于金融服务类 MCP 服务器,传输方式未标注,OAuth 授权,可直接接入 Claude、Cursor、Gemini CLI 等,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类金融服务 传输方式未标注 鉴权OAuth 授权 部署方式未标注
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

金融服务
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

financefundingremote mcp

安装与配置

未标注 OAuth 授权
上游未给出传输方式
python3 -m venv .venv && .venv/bin/pip install -e .

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
行情与财务数据

查行情、财报、汇率等金融数据。

支付与账务查询

查订单、账单与交易流水。

风控与核验

做身份、信用与风险信息核验。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude、Cursor、Gemini CLI,换客户端不用重新研究配置。

先想清楚的地方

  • 使用前要准备好OAuth 授权对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在金融服务场景里让模型直接调用外部能力的人。
  • 已经在用 Claude、Cursor 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Seiche当前 未标注 未标注 未标注 有 未标注
Alpaca MCP 62 stdio · 本地进程 支持远程部署 有 2026-05-18
OpenNews MCP 11 stdio · 本地进程 支持远程部署 有 2026-05-17
MCP Server 273df76d 未标注 stdio · 本地进程 本地部署 有 2026-05-17
tradingview-mcp 未标注 stdio · 本地进程 本地部署 有 2026-04-19
goat (Goat SDK) 未标注 未标注 未标注 无 2026-05-17

服务器档案

金融服务
字段来自项目仓库与公开索引,中文说明由本站标注
提供方beepboop2025上游标注的维护方 beepboop2025
分类金融服务本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式OAuth 授权远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · python本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

47.1 KB
来自项目仓库与上游目录的原文,未做删改

SEICHE

Release scope: 28 October 2026

Eight public research tools are available: source_rights_status, money_market_context, market_workbench, world_markets_context, gift_city_context, gold_inventory_carry, research_network, data_health.
Start with source_rights_status. Retired forecasts, backtests, editorial,
composite and derived legacy tools are unavailable for every caller while their
source reuse rights and replacements remain unqualified. A bearer token or
payment does not unlock them. Do not reconstruct withheld observations from
metadata, cached history or prior responses. Direct publisher observations keep
their own dates; restricted inputs remain explicitly metadata-only.
Authenticated callers can additionally discover five private, non-executable
Agent Room tools. Earlier feature descriptions below are historical context
where they conflict with this release scope; they do not advertise availability.

Connect and ask one question

Hosted Streamable HTTP MCP; no local server or API key for public read-only tools.
Add the products you need to Claude Code:sh claude mcp add --transport http seiche https://api.seiche.info/mcp claude mcp add --transport http liquilens https://api.liquilens.in/mcp claude mcp add --transport http undertow https://api.seiche.info/undertow/mcp

Try this first: What is the current US funding stress regime, and which dated funding events are next? Cite source dates and missing evidence.

Try it in your browser ·
Cursor configuration ·
Codex commands ·
Free three-product starter kit and skill

For a connected review, ask for institution evidence → funding context → BTC
exit estimates, with each source date, coverage gap and limitation kept separate.
Client/model costs and fair-use limits may apply. Research calls do not execute trades.

Choose the tool by the answer you need

Start with source_rights_status for feature availability and
data_health for source clocks and faults. The broader tools have different jobs:

Needed output Tool First arguments
Cross-market summary and coverage gaps world_markets_context {"section":"summary"}
Detailed rates, spreads and funding-desk evidence money_market_context {"section":"summary"}
Dated FX observations and accepted China annual histories market_workbench {"provider":"h10","base":"USD","quote":"CNY","days":365}
Dataset catalog and links to sibling-product tools research_network {"topic":"institutions","offset":0,"limit":12}
Compact cached funding metadata for an automated guard trade_safety_risk_context {}

Start with the row matching the task. The
tool selection guide includes all public tools,
example questions and a repeatable funding-review workflow.

Public US dollar funding research from source-dated Fed and Treasury observations. Live board: seiche.info.

Dated answers to common research questions explain the numbers, source dates and limits.

LiquiLens, Seiche and Undertow are connected research products from LIQUILENS PRIVATE LIMITED, a registered company in India. Built for international review workflows: institution evidence → funding conditions → market liquidity. Market coverage, source clocks and authority remain specific to each product. Company and investors · Company profile · Try the connected workflow.

> A seiche is a standing wave in an enclosed body of water — invisible from the
> shore, until it sloshes over the edge. Funding stress behaves the same way.

Seiche is a free, open source (AGPL-3.0-or-later) money-, foreign-exchange and
capital-market evidence terminal
organized around the dollar funding system —
US money markets, 22 public H.10 currency reference series, the Treasury
capital-market complex, the global basins connected through swap lines, and the
offshore-dollar crypto basin moored to the T-bill market through stablecoins. The core US board and
USD desk use free official/public APIs (FRED, NY Fed Markets, OFR STFM, Treasury
FiscalData, CFTC, ECB Data Portal, DeFiLlama, Coinbase Exchange). The global
market-pack catalog can also declare licensed or tenant sources, but public
outputs obey each source's redistribution policy and never expose restricted raw
values.

Seiche publishes a construction-PIT historical diagnostic—with misses and
eligibility limits kept visible—and is now accruing a separate as-published
forward record that can be verified instead of reconstructed. Licensed terminals
provide much broader real-time data and execution workflows, while OFR and NY Fed
dashboards provide authoritative source views. Seiche is a narrower, opinionated
research layer: forward-looking, alerting-ready, and provenance-honest. It is not
a Bloomberg/Reuters replacement, a real-time quote service, or an execution venue.

The lab

Seiche is one of three altitudes in a single liquidity lab that fills that gap:
Seiche reads the plumbing, LiquiLens organizes
institution evidence for review, and Undertow measures market liquidity itself
(who provides depth in each segment, what an exit at position size costs
today). The wiring between the three is real, not a brochure: the desk
assistant pulls the LiquiLens failure board over MCP, the MARKET tab renders
Undertow's published pack live, and the Windfetch engine reads the FETCH pack
built in the Undertow repo back into this board. On names: the internal
composite engine called "undertow" (critical slowing down,
engines/undertow.py) is unrelated to the Undertow sister product.

For a first connected review, start with
funding context in the browser.
The shared portfolio plan prioritizes a recurring
reviewer task while keeping each product's evidence and authority separate.

For a recurring agent workflow, the free starter kit
combines Seiche funding context with LiquiLens bank evidence and Undertow exit
estimates. Use the Hermes or
OpenClaw setup guide. The selected public
tools need no API key; fair-use limits and separate model-provider costs apply.

Use Seiche everywhere

Seiche is distributed as software, an MCP service, a research dataset, and
machine-readable catalog metadata. Status is receipt-based: listed means a
durable public record and receipt exist, prepared means an owner-controlled
submission or publication remains, and blocked means a named prerequisite
must clear first. “Usable in repo” describes local artifact readiness, not a
ledger status. The auditable source of truth is
distribution/submissions.csv.

Surface What is available Status
OpenBB Typed provider and obb.seiche router for funding stress, world markets, and data health (extension) 0.1.1 published on PyPI; ecosystem PR #13 archived/unmerged; official family Marketplace application submitted, review and fee waiver pending
Zenodo Release deposition metadata, citation identity, and related-source records (metadata) Prepared; no deposit or DOI claimed
Hugging Face Rights-reviewed direct-OFR dataset card and staging layout (dataset card) Validated draft; upload prepared
Kaggle Dataset metadata and reference-only staging layout (metadata) Validated draft; upload prepared
Smithery Owner-published hosted endpoint entry (live record) Listed; authenticated refresh accepted, cached aliases may lag. Public inventory rechecked 11 October: 16 tools
MCP directories Official Registry, Glama, Smithery, and eight additional live indexes (dated inventory) 11 existing records; current checks and older observations dated separately
Research notebooks Commit-pinned, hash-checking direct-OFR workflow (notebook) Usable in repo
Python / R / JavaScript Zero-secret world-markets API clients with evidence-contract checks (clients) Usable in repo
Market Atlas + structured corpus Interactive canonical market observations, rights-aware dataset receipts, full normalized BIS records, Seiche partitions, public API and MCP (contract) Live gateway; protected exports remain restricted and bulk records use bounded snapshot cursors
Docker Distroless, non-root, read-only Compose image and signed GHCR publication workflow (guide) Built and tested locally; GHCR prepared
Academic dataset 10 direct-OFR series and 11,163 audited observations, excluding restricted and derived rows (research kit) Validated draft; not submitted
Data catalogs Native-validated Croissant/Frictionless, graph-parsed DCAT 3/RO-Crate 1.3, and a DOI-free DataCite planning draft (metadata kit) Validated as labeled; publication prepared
AI integrations Hosted MCP configs for Claude Code, Cursor, VS Code, Gemini CLI, and Codex; separate OpenAI workspace/submission guidance (configs) Configs usable; OpenAI listing prepared
PyPI Python package and stdio MCP server (pip install seiche) 0.16.5 published on PyPI; use the immutable version when pinning an installation

Version 0.16.3 estuary checks the four monthly OFR money-market-fund series and six quarterly BIS series daily while preserving their observation dates and native release cadence. Signed deployment and publication acceptance are required before this source change is live.

Version 0.16.2 estuary prepares the reviewed Google-Extended policy, three dated dollar-funding answers and Riptide's public paper allocation beside the Market desk. The policy permits grounding/training only on three reviewed original explanations; source data, archives and other pages remain excluded. New signed application and publication receipts are required before these changes are live. The inherited 0.16.1 release connects the UAE DONIA reference to GIFT City's AED card and retains country funding and sovereign references for China, Japan, South Korea, Taiwan, Australia, the United Kingdom, all 21 euro-area countries and seven other European countries. Official observations retain their source dates, native frequency and gaps. The eight public MCP tools expose the country catalog through money_market_context(section="countries"); add country="JP" for a country desk or use section="india" for the India desk.
New official SOFR, EFFR, IORB, overnight reverse-repo and New York Fed funding
observations can reach the daily review without waiting through the previous
four-to-twelve-hour cache windows. Failed refreshes retain the original source
dates and stale warnings. Deployment and immutable package publication remain
subject to signed release gates.

The verified NY Fed historical importer covers 34 canonical instruments.
Archived observations retain unknown publication
clocks and actual ingestion times, preserve existing live vintages, and remain
excluded from forecast inputs that require known publication clocks. The
structured funding, forex and China workbench retains exact source dates, units,
capture evidence and missing states. Numerical China context requires its
independent signed acceptance. Verify deployment and package availability from
the release receipts and linked registries.

Eight research tools remain anonymous and free. Legacy derived tools are unavailable; client and catalog copy must preserve that boundary. See the
distribution and container trust guide for verification
and release invariants.

v2 "Deep Water" — an accountable engine fleet, layered analytics, one terminal

> v2.3 "Letters of Marque" (built in tandem across two sessions) adds the
> forecast layer and the layer that makes every other layer accountable:
> Undertow (critical slowing down — the basin's damping, measured on
> ordinary days), the Swell Forecast (the funding-stress forward curve —
> P(pop ≥ x bp) by date, six weeks out, from the public forcing calendar),
> The Stack (walk-forward ensemble of every event forecaster — rule, ML,
> analogs, Swell — plus The Tell, with a disagreement gauge), The Book
> (HELM tab — explicit daily positions on 2y/10y duration proxies, S&P and
> BTC over a T-bill base, walk-forward P&L with costs, block-bootstrap Sharpe
> CIs and mandatory benchmarks, verdict printed even when it loses), a
> hash-chained as-published track record shipped inside the static
> publish (nobody, including the operator, can quietly rewrite a bad month),
> and the Far Basin — Palimpsest's censorship-fear channel
> (palimpsest.info), a policy confession signal no market data vendor
> carries, honestly quarantined until it accrues testable history.

Engine · Question it answers
Kink Engine · Where does reserve scarcity start, and how many days away is it at the current drain rate? (live hockey-stick fit of SOFR−IORB vs reserves/GDP)
Liquidity Weather · What does the reserve path look like 6 weeks out — and which auction-settlement days land on thin ice? (TGA seasonal model + Fed drift + settlement calendar + backtested error bands)
Tail Seismograph · Are the P99 tails of SOFR/TGCR/BGCR detaching from the median — the first tell of every squeeze?
USD Money Market Desk ★ · Seven institutional-depth sections over the US cash system: policy corridor and overnight spreads; SOFR/TGCR/BGCR distributions and tails; repo-segment rates and volumes; CP−Treasury spreads; bills; liquidity buffers and Fed facilities; and MMF repo plumbing. Cross-source arithmetic uses exact common dates, repo aggregates require a fixed component set, and the descriptive NORMAL/WATCH/STRAIN/STRESS label uses a dependence-robust family-wise adjustment rather than an uncalibrated maximum.
Global Money Market Atlas ★ · A licence-aware, native-frequency catalog and available-evidence comparison across 19 registered monetary systems and 21 national sovereign-reference packs, plus a dated source-audited discovery ledger of 44 additional monetary areas across nine regions. Registration and discovery are not live coverage. Each available market is compared only with its own history and can be LIVE_REFERENCE/STALE_REFERENCE, DERIVED_CONTEXT, POLICY_ONLY, or DECLARED_UNAVAILABLE; no global stress score is manufactured from unlike rate levels.
Echo Engine · Does today's 30-day trajectory rhyme with the run-up to any historical stress episode?
Tide Tables ★ · What happened next, every time the water looked like this? Markets rhyme, so forecast like a tide table: the k nearest analogs of today's trailing state trajectory over ALL history (labeled or not, expanding-z — no look-ahead) publish their actual forward spread paths as a fan, the share followed by a funding event within 5bd (Wilson CI vs climatology), a NOVELTY gauge ("the board has never looked like this" is its own signal, and flags the fan as extrapolation), and a walk-forward hindcast that says honestly whether analogs beat the base rate.
RV X-Ray · How big is the leveraged Treasury RV complex, and what does a 5/15/30bp shock do to it?
Crowding · Where are leveraged funds most crowded relative to their own history (UST curve, SOFR/FF futures, S&P)?
Auction Digestion · Is the market choking on Treasury supply?
Warehouse · How full is the primary-dealer balance sheet — the shock absorber of last resort? (NY Fed PD stats by maturity bucket)
Resonance Engine ★ · The seiche made literal: does the same calendar forcing (month-end, quarter-end, year-end, tax dates) produce a bigger slosh than it used to? Amplification = damping loss = fragility rising while levels look calm.
Undertow ★ · The free-decay half of the resonance physics: critical slowing down (Scheffer et al.), measured continuously. Rising lag-1 autocorrelation + variance of the detrended spread/tail and a stretching recovery half-life after everyday pops = the basin losing damping on days when NOTHING is happening. Expanding percentiles only; weighted into the composite as structural evidence.
Swell Forecast ★ · A 42-business-day funding-stress forward curve: P(SOFR−IORB pop ≥ 2/5/10/20bp) by date, built from the public forcing calendar (turn/tax/settlement days each keep their full expanding distribution of historical pops), lifted by the live damping state and announced coupon settlements. It compounds to P(event by horizon), is walk-forward scored vs climatology on final-vintage history with the reliability table printed, and self-demotes to "trust the dates, not the levels" when the levels stop earning it.
Hydrophone Array ★ · How connected is the plumbing right now? (absorption ratio over 11 funding series + a live lead-lag map of which pipe is upstream)
Global Basin Coupling ★ · Are the US, euro-area, UK, India (FX channel) and crypto basins moving as one tide? Plus the global confession channel: USD swap-line draws (test operations excluded).
The Estuary / Passage ★ · Where FX settlement and physical inventory become money-market cash demand. Twenty-two H.10 currencies, advanced/emerging dollar indices, daily energy and monthly IMF commodity breadth are normalized into an upstream-pressure reading, then compared with SOFR−IORB and commercial-paper spreads. Its differentiator, The Passage, chooses each candidate target and lag on the first 60% of aligned history and calls it earned only if direction and magnitude survive the untouched final 40%; unstable stories print not earned. De-clustered analog outcomes, BIS PvP settlement structure and an editable cash-conversion lab keep statistical evidence, structural benchmarks and scenarios visibly separate. Context only, never a composite input.
Oil × Funding ★ · The barrel's balance sheet, in both directions: WTI/Brent, CP−bill and SOFR−IORB evidence; live Cushing stocks and Brent−WTI basis kept separate from dated capacity, benchmark and chokepoint references; a rolling change-on-change coupling diagnostic; the observed mechanical carry hurdle; energy/core CPI into IORB; broad foreign-official dollar-parking proxies for the recycling channel; and an editable, explicitly scenario-only lab for cargo credit, margin calls, RBI liquidity absorption and OMC commercial-paper demand. Context only, never a composite input.
Ballast ★ · The energy-futures cash ledger inside Oil × Funding. Exact CFTC contract codes for WTI and Henry Hub join weekly open interest, trader classes and paying-side concentration to public benchmark moves and contract multipliers; EIA commercial crude stocks supply the physical-collateral leg; SOFR−IORB and CP−bill show a separate funding-amplifier overlay. The output is a bounded gross mark-displacement proxy—not an observed margin call—and the headline takes the worst own-history commodity/physical percentile without blending or letting funding alone manufacture a commodity alert. Typed handoffs route exit cost to Undertow and named-institution exposure to LiquiLens. Context only, never a composite input.
Stablecoin Moorings ★ · The offshore-dollar basin's tie lines: peg deviations (USDT history + live board), total-circulation flows ($200B+ of T-bills behind them), and the 24/7 BTC canary — crypto trades when funding markets sleep.
ML Lab · Learned P(funding event within 5bd): walk-forward with a 5bd boundary embargo, benchmarked against climatology AND the rule-based index, reliability table + decision-utility scoring published. Verdict at build: ranks better than the rule (OOS AUROC 0.826 vs 0.806; 0.812 on the orthogonal feature set) but probability levels don't beat climatology — use for ranking/alerting, not literal odds. The verdict self-updates.
Station-Keeping ★ · Orbit-determination transfer: propagate the reserve system's expected state (fiscal seasonal, calendar buckets, trailing drift), CUSUM the innovation residuals, flag unmodeled "burns" — debt-ceiling cash games, RMP pace changes — often before they're narrated. Doubles as the Weather model's health monitor.
Riptide ★ · The pop prognosis — the one morning the whole desk asks the same question, answered: chop or current? Every declustered spread pop becomes a trial; the discriminators (RRP co-sign — a pop WITHOUT its mechanical quarter-end co-move is genuine scarcity, the 2025 signature; calendar bucket; damping state) feed a deliberately tiny walk-forward logistic that classifies the live pop as calendar mechanics or the start of a squeeze, with P(sticky) and P(escalates) scored pop-by-pop against the base rate on final-vintage history. Speaks only when there is a live pop; flat water is itself the reading.
The Breakwater ★ · The rescuer modeled — the feature no forecaster ships: the Fed is not weather, it is a PLAYER, and every intervention in the public record is a confession of where its pain threshold sat that day. A zero-parameter revealed-preference catalog (repo ops '19, QE '20, SRF '21, BTFP '23, QT taper '24, RMPs '25) compared with the final/current-vintage construction-PIT board for the day before each announcement yields a diagnostic threshold and a live rescue proximity gauge — which cuts both ways, and the engine says so: a forecast miss after an intervention is a save, not a false alarm.
Bathymetry ★ · The basin floor mapped from the water's motion — the physics program end to end. The daily pop statistic is treated as a diffusion and its dynamics are RECONSTRUCTED from the data (Kramers–Moyal / empirical Langevin): drift → the effective potential (the well the spread rests in, its restoring stiffness, and the escape barrier printed in units of thermal energy k_BT); the binned transition operator → the quantum-dual energy spectrum (Fokker–Planck ↔ Schrödinger: stationary density = ground state, eigenvalue moduli = energy levels, spectral gap = inverse of the slowest relaxation time — critical slowing down measured operator-theoretically, corroborating Undertow by an independent estimator); stationary probability currents → entropy production (Schnakenberg, nats/day — the arrow of time: a calm basin relaxes, a stressed one is pumped); and absorbing-boundary first passage → P(funding event within h bd · today's state) and the expected business days to the next event, Kramers' escape problem solved exactly on the measured landscape, no simulation. Expanding counts only, walk-forward scored vs climatology on final-vintage history, and the daily probability joins the Stack as its own member with its own record.
The Stack ★ · One P(funding event, 5bd) from the whole fleet: rule index, ML Lab, Tide Tables, Swell and Bathymetry calibrated per-member and blended walk-forward (with regime dummies, ~10 params on purpose). Publishes the equal-weight mean instead whenever the fitted stack fails to beat it OOS, publishes member DISPERSION — when the fleet disagrees, conviction drops — and wraps today's number in a Venn–Abers calibrated band [p0, p1] with finite-sample validity guarantees: not "our probability is 7%" but "the calibrated probability is provably between these bounds".
The Book ★ · The signal made accountable (HELM tab): a FROZEN rulebook maps the ensemble to explicit daily long/short/flat weights (2y/10y UST duration proxies, S&P 500, BTC over T-bill cash; hysteresis bands, a disagreement gate, vol targeting, per-sleeve cost haircuts), then walk-forward P&L — signal t earns returns t+1, enforced in one place and unit-tested — with stationary-block-bootstrap Sharpe CIs, Newey–West t-stats, per-episode attribution, doubled-cost rerun, and benchmarks through the identical pipeline. If it doesn't beat the static mix after costs, the page says so in bold. Every day's positions land in a hash-chained as-published ledger carried by the published site — tamper-evident by construction. Paper proxy; not advice.
Merian Modes ★ · (v2.6 "Bathysphere") The seiche eigenmodes, estimated instead of assumed. Merian's formula gives a real basin's standing-wave period from its geometry; we go the other way — Hankel-DMD (a finite-dimensional estimate of the Koopman operator: classical dynamics in the Hilbert-space clothes of Koopman–von Neumann mechanics) reads the funding basin's actual modes out of the plumbing panel: period, growth rate, current excitation. A mode with \ · λ\ · > 1 is a growing oscillation — instability visible before levels move; the ~21bd mode is the month-end forcing seen a second, independent way. The linear mode-propagation forecast is scored vs persistence and self-demotes (modes are structure, not a crystal ball).
The Gyre ★ · (v2.6) Is prediction possible at all? Takens delay embedding + empirical dynamic modeling (Sugihara): simplex-projection skill by horizon (chaos decays, noise never had skill), a phase-randomized surrogate gate for determinism beyond linear autocorrelation, the S-map θ test for state-dependent (nonlinear) dynamics, and the S-map Jacobian's local expansion rate \ · λ\ · as a live "the water is locally unstable" gauge. Tide Tables asks WHICH history rhymes; the Gyre asks whether the basin's dynamics are deterministic enough to rhyme at all.
Rogue Wave ★ · (v2.6) The tail law. Extreme value theory is literally the mathematics of rogue waves: peaks-over-threshold GPD (probability-weighted moments, bootstrap CIs, threshold-sensitivity table printed) on the SAME declustered pop statistic as PROOF. Swell's empirical exceedance curves stop dead at the largest pop in the sample; the GPD is the honest instrument for the wave that is NOT in the sample yet — the once-a-decade pop in bp, P(pop ≥ 25bp within a quarter), and whether the tail is getting heavier as the buffers drain (annual expanding ξ refits).
Far Basin ★ · The policy-fear channel: Palimpsest (palimpsest.info) measures what the Chinese state rushes to delete — the DDTI deletion-threat index, newly-targeted terms, the Generative Firewall Index — CI-published, keyless, mirrored on GitHub raw. A confession channel one basin further out, carried by no market data vendor. Honest scope: days old as a public series, so it accrues locally and stays QUARANTINED (context only, never in the composite, never a model feature) until it clears 250 daily observations.
Seiche Index · One 0–100 number with full decomposition and a regime call: CALM / EROSION / STRAIN / STRESS.

★ = methods invented for this tool.

The desk assistant: seiche ask "why is the index elevated?" (or the Ask box on
BOARD) answers strictly from a deterministic context pack of the live board — every
number cited to its engine and as-of date, "not on the board" instead of improvisation.
Routed through free-llm-router's free tiers, or any OpenAI-compatible endpoint via
SEICHE_LLM_BASE_URL; with neither configured it returns the context pack itself.

The analytics layers on top:

  • The Tell — plumbing percentile minus market-priced-stress percentile (VIX, HY/IG
    OAS, rates vol). Positive = the basin is sloshing and the screens haven't noticed.
    The whole thesis in one tradeable number.
  • The Estuary — upstream FX/material pressure minus the funding pressure already
    priced in SOFR and commercial paper. A positive Passage gap says the cash burden is
    running ahead of its usual landing zones; the discovery/holdout map then shows which
    links actually survived unseen history. Monthly physical breadth is never filled
    forward into the daily evidence, and the cash lab never pretends user assumptions
    are observed exposures.
  • The Navigator — an LLM forecaster made accountable: one committed
    P(funding event, 5bd) per data-day, grounded strictly in the live board,
    written into the hash-chained record. An LLM cannot be honestly backtested
    (it has read the history), so its FORWARD record is its only evidence and
    its weight stays zero until that record earns a hearing. seiche navigator.
  • The Communiqué — FOMC statements read as vintage-stamped data: frozen
    deterministic lexicons score policy direction, balance-sheet bias and
    funding-stress vocabulary per statement; the change vs the previous
    statement is the signal, and the Time Machine replays text as it stood.
  • The TED bridge — the ML Lab pretrains on the TED spread's 1990–2018
    funding-stress record (2008/2011/2016) in the same feature slots,
    down-weighted, and publishes the transfer gain vs the SOFR-only model
    either way.
  • The Stack + The Book — the rule index, ML Lab, Tide Tables analogs, the
    Swell curve and Bathymetry's first-passage odds all emit P(funding event, 5bd); the Stack calibrates and blends them
    walk-forward (publishing the equal-weight mean whenever the fitted blend can't
    beat it), publishes member dispersion as a first-class ambiguity signal, and
    the Book converts the result into explicit daily paper positions with costs,
    benchmarks and bootstrap CIs. Every view's daily forecast and every position is
    appended to the hash-chained PIT record — a track record no reconstruction can
    polish.
  • Turn Barometer — forecasts the next month/quarter-end turn's severity with
    leave-one-out cross-validation, always benchmarked against a naive forecast. When
    the model can't beat naive, it says so and publishes naive instead.
  • Playbook — what S&P/VIX/OAS/yields did the last N times the board looked like
    this, in native units, with n printed. Decision support, not advice.
  • PROOF — the construction-PIT diagnostic lab: the index rebuilt with expanding-window statistics
    only (causal truncation is unit-tested, but source history is final/current-vintage), recall/precision with Wilson 95%
    intervals, run-level precision (alert days are serially correlated; runs are the
    honest trials), episode-by-episode lead times including the ones it missed —
    and the orthogonal signal test: the same event-capture with the target's own
    variable family (spread/tails) removed from the signal. At build: orthogonal
    recall 69% [CI 42–87] vs 62% full, with the structural components alone at the

…(正文过长,页面只展示前面部分,完整内容见仓库)

常见问题

7 问
答案由本站依据条目字段整理,不引用上游原文
Seiche 怎么安装?

在客户端按下面的安装命令接入:python3 -m venv .venv && .venv/bin/pip install -e .。装完在客户端配置里引用同一个包或端点。

Seiche 怎么在 Claude、Cursor、Gemini CLI 里配置?

它的传输方式未标注,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Seiche 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Seiche 需要 API Key 吗?

资料标注需要OAuth 授权认证,使用前准备好对应凭据,并按最小权限配置。

Seiche 支持哪些客户端?

资料标注支持 Claude、Cursor、Gemini CLI、Codex、Zed;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Seiche 能提供哪些能力?

资料显示能力标签包括 finance、funding、remote mcp、research。具体工具清单以官方文档为准。

Seiche 需要什么运行环境?

运行时为 python。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 未标注。