ManyLives paid tools

me.manylivesv0.1.0更新於 Oct 9, 2026

67 pay-per-call agent tools: token verdicts, web extraction, crypto data, text AI, site audits.

已驗證STDIO僅桌面AI & ML Finance Web Search & Scraping

概覽

AI 產生的概覽

本機 MCP 伺服器，提供 67 個按次付費工具，涵蓋代幣判定、網頁擷取、加密資料、文字 AI 與網站稽核，以 Base 上的 USDC 付款。

功能: 提供 67 個按次收費的工具，涵蓋代幣風險判定、網頁內容擷取、加密與鏈上資料、文字 AI，以及網站或 MCP 稽核。每次呼叫以使用者自己錢包中的 Base USDC 付款，並可設定單次與每日支出上限。沒有錢包金鑰時，所有工具以免費試用方式執行，每天限 3 次呼叫。也可透過客戶端程式庫在程式碼中呼叫，並適配 Vercel AI SDK、LangChain.js 等框架。
適用情境: 適合助理偶爾需要付費資料或分析的場景，例如交易前檢查代幣、從網頁擷取欄位或執行網站稽核，且無需註冊帳號或申請 API 金鑰。免費試用適合輕度測試，常態使用則需要錢包餘額。
執行需求: 透過 stdio 在本機執行，通常用 npx 從 npm 套件 manylives-paid-tools 啟動，僅支援桌面 MCP 用戶端。無需帳號或 API 金鑰。透過環境變數 MANYLIVES_WALLET_KEY 提供持有 USDC 的專用 Base 錢包；未提供則免費試用每天 3 次呼叫。可選上限 MANYLIVES_MAX_USD_PER_CALL（預設 0.05）與 MANYLIVES_MAX_USD_PER_DAY（預設 1），以及可選的 MANYLIVES_TOOLS 清單來限制暴露的工具。需要網路存取。

安裝前請注意

每次呼叫都會從 MANYLIVES_WALLET_KEY 對應的錢包花費真實 USDC，建議使用餘額較少的專用錢包而非主錢包。支出上限按行程計算且可被調高。付款只發往說明中的 ManyLives 金庫位址。代幣判定等工具輸出屬於資料，不構成財務建議。

安裝

在 SourceWeft 中

開啟儀表板中的 ManyLives paid tools，將其新增到工作區。
為需要使用其工具的對話啟用該服務。

Desktop only，透過 STDIO。 STDIO 服務會啟動本機處理程序，因此需要 SourceWeft 桌面主機。

其他 MCP 客戶端

參照儲存庫中的啟動說明。

README

manylives-paid-tools

67 pay-per-call tools for AI agents from ManyLives: token verdicts, web page extraction, crypto and on-chain data, text AI, site and MCP audits, and more. No account and no API key. Each call costs US$0.001 to $0.05, paid in USDC on Base from your own wallet, with spend caps you set.

Try it free: without a wallet key every tool runs on the free trial (3 calls a day).

Use it from Claude, Cursor or any MCP client

Or in a JSON MCP config (Claude Desktop, Cursor, Windsurf…):

json

{  "mcpServers": {    "manylives": {      "command": "npx",      "args": ["-y", "manylives-paid-tools"],      "env": {        "MANYLIVES_WALLET_KEY": "0x… (optional: a small dedicated Base wallet holding a few dollars of USDC)",        "MANYLIVES_MAX_USD_PER_CALL": "0.05",        "MANYLIVES_MAX_USD_PER_DAY": "1",        "MANYLIVES_TOOLS": "token_verdict,web_extract_fields,site_agent_report"      }    }  }}

{  "mcpServers": {    "manylives": {      "command": "npx",      "args": ["-y", "manylives-paid-tools"],      "env": {        "MANYLIVES_WALLET_KEY": "0x… (optional: a small dedicated Base wallet holding a few dollars of USDC)",        "MANYLIVES_MAX_USD_PER_CALL": "0.05",        "MANYLIVES_MAX_USD_PER_DAY": "1",        "MANYLIVES_TOOLS": "token_verdict,web_extract_fields,site_agent_report"      }    }  }}

MANYLIVES_TOOLS is optional; leave it out to expose every tool.

Use it in code

import { createClient } from "manylives-paid-tools";
const ml = createClient({ privateKey: process.env.WALLET_KEY, maxUsdPerCall: 0.05, maxUsdPerDay: 1 });const v = await ml.call("token/verdict", { token: "0x940181a94A35A4569E4529A3CDfB74e38FD98631" });console.log(v.risk, v.flags, v.verdict);

import { createClient } from "manylives-paid-tools";
const ml = createClient({ privateKey: process.env.WALLET_KEY, maxUsdPerCall: 0.05, maxUsdPerDay: 1 });const v = await ml.call("token/verdict", { token: "0x940181a94A35A4569E4529A3CDfB74e38FD98631" });console.log(v.risk, v.flags, v.verdict);

Trading bots: check a token before you trade it (US$0.02)

const v = await ml.call("token/verdict", { token });if (v.risk === "high" || (v.market.impact_1000_usd_pct ?? 100) > 3) return skip(token, v.flags);

const v = await ml.call("token/verdict", { token });if (v.risk === "high" || (v.market.impact_1000_usd_pct ?? 100) > 3) return skip(token, v.flags);

You get contract control (owner, upgradeable, concentration), live price and the price impact of a 1,000-dollar sell, the last five minutes of transfers, trend indicators, perp funding when a market exists, red flags, a risk level and a plain-language verdict. Data, not financial advice.

Vercel AI SDK

import { tool, jsonSchema } from "ai";const tools = Object.fromEntries((await ml.tools({ only: ["token_verdict", "web_extract_fields"] }))  .map((t) => [t.name, tool({ description: t.description, inputSchema: jsonSchema(t.parameters), execute: t.execute })]));

import { tool, jsonSchema } from "ai";const tools = Object.fromEntries((await ml.tools({ only: ["token_verdict", "web_extract_fields"] }))  .map((t) => [t.name, tool({ description: t.description, inputSchema: jsonSchema(t.parameters), execute: t.execute })]));

LangChain.js

import { DynamicStructuredTool } from "@langchain/core/tools";const tools = (await ml.tools()).map((t) => new DynamicStructuredTool({  name: t.name, description: t.description, schema: t.parameters,  func: async (input) => JSON.stringify(await t.execute(input)),}));

import { DynamicStructuredTool } from "@langchain/core/tools";const tools = (await ml.tools()).map((t) => new DynamicStructuredTool({  name: t.name, description: t.description, schema: t.parameters,  func: async (input) => JSON.stringify(await t.execute(input)),}));

ml.tools() returns plain { name, description, parameters (JSON Schema), execute } objects, so any framework that takes JSON Schema tools works the same way.

Safety

Payments go only to the ManyLives treasury (0xD161874249321F8AE3000C19DdD070e46682f1B3) in USDC on Base; anything else is refused.
Calls priced above maxUsdPerCall are refused; payments stop at maxUsdPerDay (per process).
Bad input is never charged: the service checks input before the payment settles.
Use a dedicated wallet with a small balance, never your main wallet.

Full catalogue with prices and inputs: https://manylives.me/paid. MIT licence.

來源：javascript/paid-tools/README.md，提交 08057d6

工具

工具後設資料尚未被收錄。

版本歷史

v0.1.0最新Oct 9, 2026