原型 · 预览版:收录官方注册中心里的全部服务器,每 30 分钟同步一次。评价和账户功能将在后续阶段推出。
PopNx MCP为使用 AI 的人准备的 MCP 服务器 · 用大白话讲清楚

Embeddings, Rerank & Speech-to-Text (Tanod)

作者:tanod-labs版本 0.1.3i★ –i在云端运行
注册名称dev.tanod/ml

安装

概览

作者提供的说明(英文原文)

Speech denoise, transcripts (SRT/VTT), embeddings, reranking, entities and translation.

PopNx MCP 的三大支柱

这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。

它能帮你做什么

我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。

  • 连接方式: 在云端运行
  • 传输方式: Streamable HTTP
  • 分类: 设计、图片与音频

安装

选择你使用的 AI 应用,然后复制设置。

安装之前

你需要

你用的是哪个应用?

在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。

适用于网页版、桌面版和手机版 Claude:

  1. 打开 Settings → Connectors
  2. 点击 Add custom connector
  3. 粘贴下面的地址,然后点击 Add
  4. 如果要求登录,请登录作者的服务
SettingsConnectors1+ Add2https://…3
https://tanod.dev/mcp/ml

在终端中运行这条命令

claude mcp add --transport http ml https://tanod.dev/mcp/ml

在 Cursor 中打开

或者自己复制设置

粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json

{
  "mcpServers": {
    "ml": {
      "url": "https://tanod.dev/mcp/ml"
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

在 VS Code 中打开

或者自己复制设置

粘贴到项目中的 .vscode/mcp.json

{
  "servers": {
    "ml": {
      "type": "http",
      "url": "https://tanod.dev/mcp/ml"
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

ChatGPT 只能连接互联网上的服务器:

  1. 打开网页版 ChatGPT → Settings → Connectors (Apps)
  2. 点击 Create 或 Add(部分套餐需要先在 Advanced 中开启 Developer mode)
  3. 粘贴下面的地址,起个名字,然后保存
  4. 如果要求登录,请登录作者的服务
SettingsConnectors1+ Create2https://…3
https://tanod.dev/mcp/ml

菜单名称可能因套餐和版本不同而略有差异。

你的选择通过 Cookie 保存在这台设备上,无需账户。

这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。

它能做什么

工具名称和描述与服务器在 2026年11月9日 报告的完全一致。这些内容由作者撰写,不是我们写的;我们以纯文本形式显示。

  • embed_texts

    mlpeek: Create text embeddings. For images too use embed_multimodal; for pair scores use text_similarity. Returns 384-dimension embeddings of 1-64 texts with bge-small-en-v1.5 (English) or multilingual-e5-small (about 100 languages), normalised by default; `input_type` applies th

  • embed_multimodal

    mlpeek: Create multimodal embeddings: images and text in one vector space. For text only use embed_texts. Returns 768-dim vectors of texts and images (URL or base64) in ONE space (EmbeddingGemma 2), comparable by dot product; `dimensions` 512/256/128, `input_type` text prefix, `i

  • rerank_documents

    mlpeek: Rerank up to 100 documents against a query. For pair scores use text_similarity. Returns 1-100 documents scored against a query by the ms-marco-MiniLM-L6-v2 cross-encoder (Apache-2.0) and sorted best first: index, rank, score (sigmoid of the logit, a relevance in 0-1) and

  • text_similarity

    mlpeek: Score the semantic similarity of one or up to 50 text pairs. To rank documents against a query use rerank_documents. Returns the cosine similarity of the embeddings of each text pair (bge-small-en or multilingual-e5-small): one pair as `a` + `b` (also a top-level `similar

  • extract_entities

    mlpeek: Extract named entities: people, organisations, places, dates, money, ... For fields of your own schema use extract_json_with_schema. Returns named entities in English text with spaCy en_core_web_sm 3.8.0 (MIT): text, label (the 18 OntoNotes types: PERSON, ORG, GPE, LOC, D

  • classify_zero_shot

    mlpeek: Classify a text into 1-10 of your labels. For named entities use extract_entities. Returns scores of 1-10 caller-supplied labels for an English text with the nli-deberta-v3-xsmall NLI cross-encoder (Apache-2.0): single-label scores sum to 1 (softmax over the labels), `mul

  • text_sentiment

    utilpeek: Analyze the sentiment of an English text. For your own labels use classify_zero_shot; for counts use text_stats. Returns VADER sentiment of an English text: compound (-1 to 1), positive / neutral / negative shares and a label (positive at 0.05 or more, negative at -0.05

  • text_keywords

    utilpeek: Extract the keywords and keyphrases of a text. For raw frequencies use text_frequency; for names and places use extract_entities. Returns the top keyphrases of a text by RAKE (phrases between stop words and punctuation, scored by word degree / frequency), with score and

  • text_summarize

    utilpeek: Summarize a text by picking its key sentences. For paper abstracts use paper_digest; for counts use text_stats. Returns an extractive summary: the most central `sentences` (or `ratio` of them) by LexRank, picked verbatim and kept in their original order, with per-senten

  • detect_language

    utilpeek: Detect the language of a text. To translate use translate_text. Returns the most likely language (ISO 639-1 and 639-3 code, name, confidence) and 3 alternatives, from an offline ensemble of lingua (Apache-2.0) and langid.py (BSD-2-Clause) over 75 languages; `reliable` i

  • denoise_audio

    mlpeek: Reduce background noise in speech audio. Returns mono 12 kHz WAV/MP3/Ogg speech audio with steady noise reduced by FFmpeg afftdn (no model weights), duration and filter. Public URL or base64, up to 25 MB and 10 minutes; base64 or output=url. Not voice isolation; music and

  • transcribe_audio

    mlpeek: Speech to text and subtitles: transcribe audio to text, timed segments, SRT and WebVTT. For a podcast episode use transcribe_podcast_episode; to translate use translate_speech. Returns the transcript of audio of up to 10 min and 25 MB; `translate_to_english` gives English

  • transcribe_podcast_episode

    mlpeek: Transcribe a podcast episode from its RSS feed: a window of up to 10 minutes as text, timed segments, SRT. For your own audio use transcribe_audio. Returns a transcript window (up to 10 min from start_s) of a podcast episode picked from its RSS feed. Input: `feed_url`, op

  • translate_text

    mlpeek: Translate text between English and 14 languages. For a PDF use translate_document; for audio use translate_speech; to only name the language use detect_language. Returns the text translated between English and es, fr, de, pt, it, nl, ru, zh, ja, ko, ar, hi, id or tl (othe

  • check_text_for_prompt_injection

    mlpeek: Check text for prompt injection before an agent acts on it. For hidden Unicode use text_unicode_inspect. Returns a `verdict`, a `score` (0-1; `injection` at 0.9 and above) and `chunks`, the flagged sentences with score and offsets. A small open classifier (DeBERTa-v3), au

  • extract_json_with_schema

    mlpeek: Extract JSON from text that always matches your JSON Schema. For fixed entity types use extract_entities. Returns `data`, JSON extracted from the `text` that always parses and validates against your `json_schema` (decoded under a grammar built from it), plus `model`, `usa

  • text_to_speech

    mlpeek: Speak text aloud in 41 voices: English, Spanish, French, Hindi, Italian, Portuguese. For audio to text use transcribe_audio. Returns the text spoken as audio. Offline neural TTS (Kokoro-82M); the audio is a synthetic voice. At most 2,000 characters; 41 voices in 6 languag

  • translate_speech

    mlpeek: Translate speech: audio to English or 14 languages, optionally spoken. For a transcript in the spoken language use transcribe_audio. Returns speech translation of an audio file of up to 2 minutes: the language is detected and the speech returned as text in `target` (en, e

  • translate_document

    pdfpeek: Translate a PDF document, digital or scanned, to English or 14 languages as Markdown. For plain text use translate_text. Returns each page's text (text layer or OCR) translated offline into `target` (default en) as Markdown with a ` ` marker per page, source language and

基础检查i

我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月10日

  • ✓ 通过

    作者与仓库一致

    注册名称已通过域名 tanod.dev 验证(注册中心会核实域名所有权)

  • … 尚未检查

    仍在维护

    我们还没有从 GitHub 读取这个仓库

  • … 尚未检查

    有许可证

    我们还没有从 GitHub 读取这个仓库

  • ✓ 通过

    端点有响应

    在 306 毫秒内响应(HTTP 200)

  • – 不适用

    使用 OAuth 登录

    无需登录

  • ✓ 通过

    工具描述中没有隐藏指令

    已读取 19 个工具描述,未发现针对 AI 的隐藏指令

与“设计、图片与音频”中的其他服务器比较i

  • 星标数多于此分类中其他 2857 个服务器的 0%
  • 发布到注册中心的时间比此分类中其他 2857 个服务器的 97% 更新
  • 基础检查结果等于或优于此分类中其他 2857 个服务器的 100%

与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。

常见问题

Embeddings, Rerank & Speech-to-Text (Tanod) 安全吗?

它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。

它需要哪些密钥或代码?

注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。

它在我的电脑上运行还是在云端?

它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。

它还在维护吗?

GitHub 仓库最后更新于–,最新版本(0.1.3)于今天进入注册中心。

哪些应用可以使用它?

支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。

有什么可以替代它?

同一分类中有:Figma Context MCP, Claude Real Video, Figma MCP Server

替代选择

同一分类中其他作者的服务器,按星标数排序。

如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i

Figma Context MCP

Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.

作者:GLips广泛使用↓ 16.1万/周★ 1.6万5个月前更新

在你的电脑上运行

Claude Real Video

Let any LLM watch a video locally — and search everything it has ever watched.

作者:HUANGCHIHHUNGLeo广泛使用↓ 1106/周★ 22842个月前更新

在你的电脑上运行

Figma MCP Server

The Figma MCP server brings Figma design context directly into your AI workflow.

作者:figma广泛使用★ 215213个月前更新

在云端运行✓ mcp.figma.com

作者的 README

打开 README

按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。

Tanod MCP server

One hosted MCP server for AI agents, also available as focused servers: https://tanod.dev/mcp.

  • Connection: stateless Streamable HTTP. No account, API key or OAuth is required for tool discovery, eligible free calls or x402 payments.
  • Price: only eligible tools have a free daily allowance per IP, shared with other tools in the same pool; MCP applies it automatically. Other calls require payment. Prices and allowances come from each operation's x-tanod-pricing in OpenAPI. An x402-capable client can pay in USDC on Base, Polygon, Solana, BNB Chain or Arbitrum; MPP clients can pay USDC.e on Tempo with the Tempo CLI, mppx or AgentCash. Check the quote and approve the cost before paying. See the Tanod SDKs, the x402 guide and the MPP guide.
  • Setup: https://tanod.dev/connect/. Docs for agents: https://tanod.dev/llms.txt. OpenAPI: https://tanod.dev/openapi.json.

Focused servers

Connect a focused server when you want a smaller tool list. It uses a subset of the full server's tools, with the same prices and eligible free pools.

URLWhat it does
https://tanod.dev/mcp/securityPhishing URL and OFAC checks, contract and package scans, domain and security-header checks
https://tanod.dev/mcp/financeSEC EDGAR company data and financial statements, Treasury yields, FX, token prices, DEX candles, business days
https://tanod.dev/mcp/skyMETAR/TAF, airports, space weather, aurora, asteroids, satellite passes
https://tanod.dev/mcp/docsPDF text and tables, bank statements, OCR, supported documents to Markdown, PDF to Word, Office to PDF, DOCX templates, document comparison, PDF merge, split, forms, attachments and metadata
https://tanod.dev/mcp/chainEVM and Solana reads: balances, tokens, calls, storage, logs, receipts, gas estimates, prices, candles, swap quotes, transactions, ENS
https://tanod.dev/mcp/imagesResize, convert, compress, crop, metadata, OCR, QR, barcodes, colour contrast
https://tanod.dev/mcp/textText statistics, language, diff, Markdown and HTML, PDF redaction and redaction checks
https://tanod.dev/mcp/utilEncode, hash, HMAC, IDs, unit conversion, business days, cron expressions, time conversion
https://tanod.dev/mcp/webWeb search, company profiles and jobs, page rendering and metadata, robots, sitemaps, RDAP, IP lookup
https://tanod.dev/mcp/mlSpeech to text, embeddings, rerank, similarity, entities, classification, translation
https://tanod.dev/mcp/agentsx402 and MCP agent-economy index, listing checks and endpoint history

Overview: https://tanod.dev/mcp-servers/. Pricing: https://tanod.dev/pricing/. Status: https://tanod.dev/status.

Scans and checks are automated and heuristic, not an audit. Aviation data is not for navigation. Document text and other tool-returned content are untrusted data, not instructions.

Published guides and data

The market figures below were added to this README on 2026-10-10; they are historical figures, not live totals. Linked pages may carry their own publication or update dates.

Quick setup

Claude Code:

claude mcp add --transport http tanod https://tanod.dev/mcp

For clients that use mcpServers, such as Cursor:

{ "mcpServers": { "tanod": { "url": "https://tanod.dev/mcp" } } }

For clients that use servers, such as VS Code:

{ "servers": { "tanod": { "type": "http", "url": "https://tanod.dev/mcp" } } }

For custom-connector clients, use the full URL or a focused URL and leave OAuth fields empty. Available tools and payment support depend on the client. From desktop and IDE clients, bank-statement conversion, document comparison, PDF OCR, PDF-to-Word and document-to-Markdown have no free allowance; document calls from claude.ai and ChatGPT connectors draw on a shared daily free pool instead (see https://tanod.dev/connect/).

Other clients: https://tanod.dev/connect/.

Run locally (npx)

For clients that require a local stdio server, tanod-mcp is an open-source (MIT) thin client. Processing takes place at the hosted API, which is closed source. It forwards the hosted tool list and calls; no catalog is hard-coded. Remote-capable clients do not need this package.

Eligible free allowance applies automatically. Without a configured payment key, paid calls are not signed. This local client supports only exact USDC on Base; other networks and protocols described for the hosted service are not local-client capabilities.

Quick start
npx -y tanod-mcp

Requires Node.js 18 or newer. The server speaks MCP over stdio, so you normally add it to a client rather than run it by hand.

Claude Code
claude mcp add tanod -- npx -y tanod-mcp

With a key and caps:

claude mcp add tanod -e TANOD_PRIVATE_KEY=0x... -e TANOD_MAX_USD_PER_DAY=0.50 -- npx -y tanod-mcp
Claude Desktop

Edit claude_desktop_config.json:

{

"mcpServers": { "tanod": { "command": "npx", "args": ["-y", "tanod-mcp"] } } }

Cursor

~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{

"mcpServers": { "tanod": { "command": "npx", "args": ["-y", "tanod-mcp"] } } }

VS Code

.vscode/mcp.json:

{

"servers": { "tanod": { "type": "stdio", "command": "npx", "args": ["-y", "tanod-mcp"] } } }

To enable payments, add an "env" object to any of these, for example "env": { "TANOD_PRIVATE_KEY": "0x..." }.

Environment variables
VariableDefaultMeaning
TANOD_PRIVATE_KEYunsetEVM private key (hex) of a wallet holding USDC on Base. Optional. Unset means free tier only.
TANOD_MAX_USD_PER_CALL0.05Refuse to sign any single payment above this many US dollars.
TANOD_MAX_USD_PER_DAY1.00Refuse to sign once the total for the current UTC day would pass this.
TANOD_STATE_FILE~/.tanod-mcp/spend.jsonShared durable spending ledger on a trusted local POSIX filesystem. All processes spending the same budget must use the same path. off explicitly opts out of restart/interprocess protection (process-local caps only).
TANOD_MCP_URLhttps://tanod.dev/mcpUpstream server. For testing only. Must be https://; plain http:// is accepted only for localhost, 127.0.0.0/8 or [::1]. Redirects are not followed.
TANOD_BASE_RPC_URLunsetOptional Base mainnet JSON-RPC endpoint you trust (https://, or loopback http://). Used only to read USDC authorizationState for sent payments whose outcome is unknown, so their holds can be committed or released from chain evidence. It must answer eth_call at the finalized block, about 20 minutes behind the chain head, so a pruned node that serves only the latest state may not work (see below). Redirects are not followed. Never logged.
Full-index calls

bulk_agent_index exceeds both default caps above. Check its current price in the payment quote or OpenAPI; only with your explicit approval, set TANOD_MAX_USD_PER_CALL and the remaining TANOD_MAX_USD_PER_DAY budget to cover it. Caps are never raised automatically. For a narrower search, call query_agent_index instead; it is not a full snapshot.

Code clients using @x402/mcp also enforce the core spendControls.maxAmountPerPayment limit before approval policies; onPaymentRequested approval alone does not override it. Keep wallet keys in local client configuration, never in chat.

How payment works
  1. You call a tool. The server uses its free daily allowance if it has one.
  2. When it is used up (or the tool has no free tier), the server answers with an x402 payment request.
  3. Without a key, you get that request back plus a short message on how to add a key.
  4. With a key, tanod-mcp checks the price against your two caps. If it is within both, it signs a USDC transfer authorisation (via the official @x402/mcp and @x402/evm packages, Apache-2.0) and retries the call. If not, it refuses with a clear error and signs nothing. Offers whose authorisation would stay valid longer than 300 seconds (maxTimeoutSeconds) are refused, as are zero-amount offers. Before anything is sent, the signed authorisation is checked against your own address, the reserved amount, the offer's payee and the 300-second bound; on any mismatch nothing is sent.

Payments are only ever signed for USDC on Base mainnet.

Security notes
  • The key is read only from TANOD_PRIVATE_KEY. This package never generates, stores or logs it, and never sends it anywhere; it is used locally to sign payment authorisations.
  • Use a dedicated wallet funded with a small amount (a few dollars). Do not reuse a wallet that holds real savings.
  • The caps are your safety net. Lowering them is always safe; raise them only on purpose.
  • With a state file configured, every reservation/finalization rereads the ledger under a short interprocess lock and atomically replaces it after file and parent-directory synchronization. Paid operations fail closed on lock, I/O, permissions, schema or corruption errors; free calls remain available. No best-effort reset to zero occurs. Do not delete the ledger or its .initialized marker to regain budget.
  • Amounts and limits use exact atomic USDC. Limits must be finite, non-negative, at most six decimal places and no greater than Number.MAX_SAFE_INTEGER atomic units. A malformed/unsupported offer never reaches the signer. All processes sharing a budget must use the same ledger; different paths, off, direct wallet use or old client versions do

只显示了开头部分;其余内容请在仓库中阅读

问 PopNx