Skylit

ai.skylit由 skylit.ai 发布官方发布者v0.1.1更新于 Oct 24, 2026

Options flow, volatility, dealer positioning (GEX) and dark pool data for AI trading agents.

已验证Streamable HTTP可网页运行金融数据与分析

概览

AI 生成的概览

Skylit 是一个托管式只读 MCP 服务器，让 AI 助手获取美股期权流、波动率、做市商持仓（GEX）与暗池数据。

可以做什么
  • 检索与发现期权流：流搜索、活跃标的、到期日、流列表、交易与聚合评分、扫单与动量。
  • 筛选并比较合约与标的：日/周榜单、异常成交量与未平仓量、多空与压力比率、Vol/OI、价内外程度、统计与历史。
  • 查看期权链、图表与相对成交量，以及全市场与板块的资金流、广度与潮汐。
  • 读取暗池成交与最大单笔成交。
  • 查询 gamma/vanna 热力图（Heatseeker）以及 IV、期限结构、锥形、曲面、偏斜与事件等波动率套件（Tempest）。
  • 获取一次调用即得的情报摘要（关键价位、波动率背景、变化对比、市场简报、资金流解读），并查看账户用量与积分余额。
适用场景
适合助手需要回答美股期权持仓、资金流、隐含波动率或暗池活动相关问题时使用，例如某标的的关键 gamma 价位或当日资金流基调。更适合研究与分析型代理，而非下单执行。
运行要求
远程 Streamable HTTP 端点（mcp.skylit.ai），无需本地安装。需要具备 API 访问权限的 Skylit 账户（测试版）：可通过 Skylit OAuth 登录，或在 Authorization 头中提供 Bearer API 密钥（来自 Skylit 开发者页面）。调用会消耗付费 API 积分，部分 Tempest 工具取决于套餐。
安装前请注意
调用会消耗 API 积分（1 积分 = 0.001 美元，单次通常 1 至 5 积分），代理使用会产生费用；失败调用会退还。API 密钥为个人所有，须妥善保密。工具均为只读、不会下单，但数据仅授权用于个人交易、研究与教育；商业使用或再分发需书面许可。数据为市场信息，不构成投资建议。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Skylit，将其添加到工作区。
  2. 连接 所需的 凭据。它们会被私密存储，绝不会在公开页面上展示。
  3. 为需要使用其工具的对话启用该服务。

Web executable，通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体，或 SourceWeft 对话中。智能体会读取该服务器的安装指南，先向你展示发布者、端点和登录方式，经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/ai.skylit%2Fmcp/install.md，并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前，先告诉我它的发布者、端点和登录方式，等我确认。不要把 API 密钥写进命令或我会提交的文件；问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。 客户端会在首次连接时完成 OAuth 登录。

{
  "mcpServers": {
    "skylit": {
      "type": "http",
      "url": "https://mcp.skylit.ai/mcp"
    }
  }
}

README

Skylit MCP

Skylit is the data intelligence platform for AI trading: options flow, volatility, dealer positioning and dark pool data for US markets. The Skylit MCP server gives any MCP client (Claude, ChatGPT, Cursor, VS Code, Gemini CLI, Codex and others) read-only tools over that data.

This repository holds setup guides for each client, the Claude Code plugin, a Gemini CLI extension, an agent skill, and runnable examples. It contains no server code; the server is hosted by Skylit.

Status: beta. The API and MCP server are open to Skylit members with API access. Check yours on the Developer page.

Quick connect

URLhttps://mcp.skylit.ai/mcp
TransportStreamable HTTP
AuthSign in with Skylit (OAuth): add the URL and approve on the Skylit page. Or send Authorization: Bearer <API key> with a key from the Developer page.

Claude Code:

bash
claude mcp add --transport http skylit https://mcp.skylit.ai/mcp# then /mcp -> skylit -> Authenticate

Cursor, LM Studio, Warp and other clients that use an mcpServers JSON file:

json
{  "mcpServers": {    "skylit": {      "url": "https://mcp.skylit.ai/mcp",      "headers": { "Authorization": "Bearer YOUR_SKYLIT_API_KEY" }    }  }}

Leave out headers to sign in with Skylit instead. Field names differ between clients (url, httpUrl, serverUrl, servers), so check your client below.

Then ask: "What are the key gamma levels and today's flow tone for SPY?"

Python

bash
pip install skylit
python
from skylit import SkylitSkylit().gex_levels("SPY")   # reads SKYLIT_API_KEY

Source and docs: packages/python.

TypeScript

bash
npm install @skylitai/sdk
ts
import { Skylit } from "@skylitai/sdk";await new Skylit().gexLevels("SPY");   // reads SKYLIT_API_KEY

Source and docs: packages/typescript.

Staying current

This repo is checked against the live API every 6 hours (scripts/sync.py, sync workflow): the tool catalog and REST specs in reference/ are refreshed from Skylit's published docs, and any guide, skill, example or SDK that names a tool or endpoint that no longer exists fails the check and opens an issue.

Clients

ClientSign in with SkylitAPI keyGuide
Claude (claude.ai, Desktop, mobile)YesDesktop, via mcp-remoteclients/claude.md
Claude CodeYesYesclients/claude-code.md
CursorYesYesclients/cursor.md
VS Code / GitHub CopilotYesYesclients/vscode.md
WindsurfYesYesclients/windsurf.md
ClineNot documentedYesclients/cline.md
ZedYesYesclients/zed.md
Gemini CLIYesYesclients/gemini-cli.md
Codex CLIYesYesclients/codex-cli.md
ChatGPT (developer mode)YesNoclients/chatgpt.md
GooseYesYesclients/goose.md
LM StudioYesYesclients/lm-studio.md
RaycastYes (untested)Yes (untested)clients/raycast.md
ContinueNot documentedYesclients/continue.md
WarpYesYesclients/warp.md
PerplexityYes (unverified)Yes (unverified)clients/perplexity.md
Any other MCP client, mcp-remote, MCP Inspector, raw HTTPYesYesclients/generic.md

"Sign in with Skylit" and "API key" reflect each client's own documentation as of 2026-09-30. Guides mark anything we could not confirm.

Intelligence toolset

One-call answers for agents, each with a short factual summary, compact data and meta (cost, balance, asOf, partial). They have their own tool list, so an agent loads only these:

https://mcp.skylit.ai/mcp?toolset=intelligence
ToolAnswersCredits
explain_levelsKey dealer-positioning levels for a symbol1
vol_contextImplied volatility, IV rank, expected moves, scheduled events2
whats_changedHow the positioning board moved since a given time6
market_briefVIX complex, market-wide options flow, SPX and QQQ levels5
flow_contextToday's options flow read for a ticker9

Details: Intelligence tools.

Tools

69 tools, all read-only. Most wrap one Skylit REST endpoint with the same credit cost; the intelligence tools combine several and are served on their own list (/mcp?toolset=intelligence). Full catalog with arguments and prices: www.skylit.ai/docs/mcp/tools; machine-readable copy: reference/tools.json.

GroupTools
Discoveryflow_search, list_active_underlyings, expirations
Scores & tradesflow_feed, trade_score, aggregate_score, flow_aggregate
Sweeps & momentumsweeps, flow_momentum, flow_baseline
Strike & tide concentrationflow_strikes, flow_tide, by_strike
Screenerstop_underlyings_daily, top_underlyings_weekly, top_contracts_daily, top_contracts_weekly, unusual_volume, unusual_oi
Bull/bear & pressure ratioschain_bull_bear, contract_bull_bear, chain_ratio, contract_ratio
Stats, Vol/OI & moneynessunderlying_stats, underlying_bulk_stats, contract_bulk_stats, underlying_history, contract_history, flow_historical_compare, contract_stats, vol_oi, moneyness
Chains, charts & RVOLoption_chain, underlying_chart, contract_chart, underlying_rvol, contract_rvol, flux_field
Market-wide & sectormarket_overview, market_tide, market_breadth, sector_flow
Dark pooldark_pool_trades, dark_pool_top_prints
Heatseeker — gamma/vanna heatmapsheat_heatmap, heat_levels, heat_historical_heatmap, heat_stats_daily, heat_symbols
Tempest — volatility suitetempest_iv, tempest_term, tempest_cones, tempest_sigma, tempest_surface, tempest_tilt, tempest_events, tempest_snapshot, tempest_market, tempest_screener, tempest_history, tempest_derived, tempest_status, tempest_symbols
Accountaccount_usage
Intelligence toolsexplain_levels, vol_context, whats_changed, market_brief, flow_context

Tempest tools depend on your plan; without access they return not_entitled at no charge. OHLCV price bars (Atlas) are available over REST, not MCP.

Pricing

  • Calls spend API credits: 1 credit = $0.001. A typical call costs 1 to 5 credits.
  • Failed calls are free: any 4xx or 5xx is refunded.
  • account_usage is free and returns your balance and limits. Each tool result's meta includes the remaining balance.
  • Per-tool costs: tool catalog. Plans and included credits: www.skylit.ai/pricing.

REST API

The same data is available over REST with the same key. Each host serves its OpenAPI spec at /v1/openapi.json:

HostData
https://api.skylit.aiDealer positioning (Heatseeker), volatility (Tempest), options flow and dark pool (Flowseeker), account
https://flow-api.skylit.aiOptions flow and dark pool (alias for the Flowseeker paths)
https://atlas-api.skylit.aiOHLCV price bars (Atlas)

See examples/python/rest-quickstart and the API reference.

What's in this repo

PathPurpose
clients/Setup guide per MCP client
plugins/claude-code/ and .claude-plugin/marketplace.jsonClaude Code plugin: /plugin marketplace add SkylitAI/skylit-mcp, then /plugin install skylit@skylit
gemini-extension.json and GEMINI.mdGemini CLI extension: gemini extensions install https://github.com/SkylitAI/skylit-mcp
skills/skylit/Agent skill on using the tools well: npx skills add SkylitAI/skylit-mcp
examples/Research agents: OpenAI Agents SDK, LangChain/LangGraph, Claude Agent SDK, Vercel AI SDK, plain REST
server.jsonOfficial MCP Registry entry (ai.skylit/mcp)
llms-install.mdInstall steps for AI coding agents (Cline)

The examples are research agents. They read data and never place orders; order execution is left to your own broker.

Data and licensing

  • This repository (code and docs) is MIT licensed. See LICENSE and NOTICE. The license does not cover Skylit data.
  • Skylit data is governed by the API Terms. Keys are personal. Using data for your own trading, research and education, including private agents that act for you, is covered. Commercial use, redistribution and republishing bulk data need a written license: contact [email protected].
  • Skylit data is market information, not investment advice.

Links

来源：README.md，提交 b720b03

工具

0
该服务器需要登录，因此 SourceWeft 无法在没有账号的情况下读取其工具，也未收录。已连接账号所能看到的内容可能因账号及其权限而异。

版本历史

1
  1. v0.1.1最新Oct 1, 2026

评分

已满足适用检查项的 85%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通必需25 分
  • 已声明认证方式必需10 分
  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分