DeepStack: Independent analysis of the AI economy

Where informed investors disagree, the evidence each side stands on, and the date that settles it.

我该使用它吗

质量与安全性

A
描述质量
100%
模式完整度
77%
命名质量
80%
投毒风险
100%
权限匹配度
100%
协议合规性
100%

基于对工具定义和协议合规性的自动分析。

上下文开销

~1,649token 数(工具定义)
~1.0 KB典型响应大小
对注意力有中等影响(占 128k 上下文窗口的 1.29%)

这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高,可用于其他任务的注意力就越少。

安装

一键安装

将以下内容添加到你的 `claude_desktop_config.json` 文件中:

{
  "mcpServers": {
    "deepstack": {
      "url": "https://mcp.deepstack.ltd/mcp"
    }
  }
}

远程端点

https://mcp.deepstack.ltd/mcpstreamable-http

它能做什么

工具清单

工具(8)

🟢 只读🟡 写入🔴 删除⚪ 未知
🟢deepstack_search(query, limit, kinds)

Use when the user asks where informed investors disagree on a topic of the AI economy, or wants DeepStack’s published analysis on a company, a person or a question; returns the matching stories, sides, tests and entities with their canonical URLs, or answers “DeepStack has no formed side on …” with up to three nearest items, labeled as nearest. Use it first when the slug or name is not known.

输入模式

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Words to look for: a company, a person, a theme, a figure."
    },
    "limit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 20,
      "default": 8,
      "description": "How many hits to return."
    },
    "kinds": {
      "type": "array",
      "items": {
        "type": "string",
        "enum": [
          "story",
          "question",
          "sides",
          "company",
          "person",
          "test",
          "term",
          "layer",
          "thesis"
        ]
      },
      "description": "Restrict the hits to these kinds; all kinds by default."
    }
  },
  "required": [
    "query"
  ],
  "additionalProperties": false
}
🟢deepstack_story(slug, part, detail)

Use when the user wants one Part of The Buildout read in full, by slug or part number: headline, dek, The DeepStack Call (stance, conviction, three actions, what would change the call, the side it takes and the case against it, the next dated test), key numbers, the full text as plain paragraphs, the dated tests, any result, the companies and people named, and the sources, each with its URL.

输入模式

{
  "type": "object",
  "properties": {
    "slug": {
      "type": "string",
      "description": "The story slug, as in https://deepstack.ltd/stories/<slug>/."
    },
    "part": {
      "type": "integer",
      "minimum": 1,
      "description": "The part number in The Buildout, as an alternative to the slug."
    },
    "detail": {
      "type": "string",
      "enum": [
        "full",
        "summary"
      ],
      "default": "full",
      "description": "“summary” leaves out the body paragraphs."
    }
  },
  "required": [],
  "additionalProperties": false
}
🟢deepstack_sides(slug, part)

Use when the user asks for the bull and bear case, who disagrees with a thesis, the counter-argument, or what would change a side’s mind. Returns, for one story by slug or part number, The DeepStack Call and its side, every attributed, paraphrased voice on each side with its date and source link, the strongest case against each side, where voices cross, and the dated tests that settle it.

输入模式

{
  "type": "object",
  "properties": {
    "slug": {
      "type": "string",
      "description": "The story slug."
    },
    "part": {
      "type": "integer",
      "minimum": 1,
      "description": "The part number, as an alternative to the slug."
    }
  },
  "required": [],
  "additionalProperties": false
}
🟢deepstack_docket(window_days, include_past, include_signal, status, part)

Use when the user asks what will decide a dispute, what to watch, or the next dated test. Returns the dated tests that will settle each story, from today forward: date and precision, status, the event, the resolver (which series, release or earnings print decides it) and the story, each with its URL.

输入模式

{
  "type": "object",
  "properties": {
    "window_days": {
      "type": "integer",
      "minimum": 1,
      "maximum": 730,
      "default": 120,
      "description": "How far ahead to look."
    },
    "include_past": {
      "type": "boolean",
      "default": false,
      "description": "Also list tests dated before today."
    },
    "include_signal": {
      "type": "boolean",
      "default": false,
      "description": "Also list the tests that have no date and resolve on a signal."
    },
    "status": {
      "type": "string",
      "enum": [
        "all",
        "pending",
        "confirmed",
        "refuted",
        "inconclusive"
      ],
      "default": "all"
    },
    "part": {
      "type": "integer",
      "minimum": 1,
      "description": "Only the tests of one part."
    }
  },
  "required": [],
  "additionalProperties": false
}
🟢deepstack_results

Use when the user asks what resolved and how the Call fared. Returns the tests that came in: verdict, what happened with the primary source and the number, what it did to the Call, and the rule that no hit rate is shown until ten results exist.

输入模式

{
  "type": "object",
  "properties": {},
  "required": [],
  "additionalProperties": false
}
🟢deepstack_entity(slug, ticker, name, kind)

Use when the user asks what the voices argued about a company, or what a person argued, in The Buildout; by slug, ticker or name. Returns where it appears, the figures the stories reported, each argument attributed, dated and linked, and for a listed company the delayed market snapshot with its source. Never a score or a ranking.

输入模式

{
  "type": "object",
  "properties": {
    "slug": {
      "type": "string",
      "description": "The entity slug, as in https://deepstack.ltd/companies/<slug>/ or /people/<slug>/."
    },
    "ticker": {
      "type": "string",
      "description": "A listed company’s ticker, for example NVDA."
    },
    "name": {
      "type": "string",
      "description": "The name, when the slug is not known."
    },
    "kind": {
      "type": "string",
      "enum": [
        "company",
        "person",
        "organization"
      ],
      "description": "Narrow the lookup to companies, to people or to the organizations that contribute argument (listed with the people)."
    }
  },
  "required": [],
  "additionalProperties": false
}
🟢deepstack_tape

Use when the user asks for the macro figures DeepStack tests its notes against, each with its producer and date: Treasury yields, the real yield, the S&P 500, VIX, Brent, US natural gas, copper, the Fed funds target, CPI and payrolls, with the weekly change and the as-of time of each series.

输入模式

{
  "type": "object",
  "properties": {},
  "required": [],
  "additionalProperties": false
}
🟢deepstack_counter(claim, limit)

Use when the user states a thesis or a claim about the AI economy and wants the strongest case against it. Returns the side the claim sits on (or the question, when no side label fits), the strongest case against that side from a voice outside it, dated and linked, more opposing voices, and the dated tests that settle it; below the relevance floor, that DeepStack has no formed side on the claim.

输入模式

{
  "type": "object",
  "properties": {
    "claim": {
      "type": "string",
      "description": "The thesis as the user states it, for example “AI demand is real”."
    },
    "limit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 5,
      "default": 3,
      "description": "How many further opposing voices to return after the strongest case."
    }
  },
  "required": [
    "claim"
  ],
  "additionalProperties": false
}

社区

评价此服务器

证据

最近观测

已验证未记录版本8 个工具