Skip to main content
Glama
Alpha Sentinel Market Intelligence MCP

Alpha Sentinel Market Intelligence MCP

Alpha Sentinel - Market Intelligence x402 MCP Server

⭐ 链上赚取奖励的预测智能代理! ★

一个生产级 x402 市场服务,通过付费 MCP 工具为 AI 智能体提供实时加密市场监控、波动性警报和情绪分析。

🎯 任务

通过处理微支付 API 端点和 MCP 工具,提供具有 <30s 延迟的可操作实时市场情报。成功状态 = 可操作警报,每次调用 ROI 可衡量!

Related MCP server: Enterprise Crypto MCP Gateway

💰 定价

根据 x402 市场调研(中位价格为 $0.014,约 90% 低于 $0.10):

资源

价格

配额(专业版)

描述

price_feed()

$0.005

5000/月

实时价格查询

volatility_alerts()

$0.02

500/月

异常检测 + 阈值

sentiment_analysis()

$0.01

1000/月

社交情绪聚合

risk_assessment()

$0.03

300/月

多因素风险评分

market_report()

$0.15

50/月

综合每日报告

Tool Credit Pack

$1.00

100 calls

灵活积分包

免费层级: 每月 50 次调用,每分钟 5 次限流
专业层级: 无限配额,每分钟 120 次,优先结算

🛠️ MCP 工具

核心智能工具

  1. fetch_price() ⭐ 包含免费层级

    • 从 CoinGecko API 获取最新价格

    • 支持:BTC、ETH、前 100 种山寨币

    • 响应:{ symbol, price_usd, change_24h, volume }

  2. analyze_volatility()

    • 通过 Z-score 统计分析检测价格异常

    • 可配置阈值(默认:2σ)

    • 返回异常分数 + 警报级别

  3. aggregate_sentiment()

    • 从 X/Twitter、Reddit、Telegram 来源聚合

    • 基于 NLP 生成情绪评分(-1.0 至 +1.0)

    • 趋势检测与动量指标

  4. calculate_risk_score()

    • 多因素风险评估(波动性、流动性、相关性)

    • 风险等级:LOW/MEDIUM/HIGH/CRITICAL

    • 包含置信区间

  5. generate_market_report()

    • 综合每日/每周/每月报告

    • 整合所有数据源

    • PDF + JSON 输出格式

📦 +代表付费组合产品

  • 美国租赁尽职调查包(依赖城市合规集成层):$1.50

    • 涉及租赁许可证、违规单和租户权利

    • 多物业批量处理

    • 市政位置开放数据关联

  • 脉冲快讯:$0.05

    • 基链 + 市场指标综合分析

    • 12 区块采样深度

    • 实时现货价格集成

  • 信用决策费:$0.01

    • 低成本交易价查询层,用于常规交易前检查

    • 主网交易前即时反馈

🏗️ 架构

┌─────────────────────────────────────────────────────┐
│           Alpha Sentinel MCP Server                  │
├─────────────────────────────────────────────────────┤
│                                                      │
│  [MCP Layer]                                         │
│    ├─ FastMCP Server (stdio + SSE)                  │
│    └─ Tool Registry (SSOT)                          │
│                      │                              │
│  [Commerce Layer]                                    │
│    ├─ x402 Middleware (verify+settle)               │
│    ├─ Quota Management (free/pro tiers)             │
│    ├─ Challenge Cache (sync, 10s TTL)               │
│    └─ Stripe Integration (fiat rail fallback)       │
│                      │                              │
│  [Intelligence Layer]                                │
│    ├─ Price Feed (CoinGecko API)                    │
│    ├─ Sentiment Engine (X/Reddit/NLP)               │
│    ├─ Volatility Analyzer (statistical models)      │
│    └─ Risk Calculator (ML-enhanced)                 │
│                      │                              │
│  [On-Chain Layer]                                    │
│    ├─ CDP Facilitator (Base mainnet)                │
│    ├─ Ledger (spend/revenue tracking)               │
│    └─ Swarm Agency (buy-compose-resell)             │
│                                                      │
└─────────────────────────────────────────────────────┘

🚀 快速开始

前置条件

  • Python 3.12+

  • Redis(可选,用于生产环境)

  • x402 卖家凭证(与买家凭证分开!)

  • CoinGecko API 密钥(免费层级即可)

初始设置

# Clone and setup
git clone https://github.com/kwizzlesurp10-ctrl/alpha-sentinel-mcp
cd alpha-sentinel-mcp
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Configure environment
cp .env.example .env
nano .env  # Edit with your credentials

# Run locally
make up  # API (8403) + dashboard (5174)

环境变量

# Seller (receive revenue)
X402_PAY_TO_ADDRESS=0xAB745e5F...  # Separate from buyer!

# Buyer (hot wallet for spending)
EVM_PRIVATE_KEY=<your_buyer_key>    # NEVER on Render!

# APIs
COINGECKO_API_KEY=<your_key>
X_API_KEY=<twitter/x_api_key>       # Optional

# Commerce
STRIPE_SECRET_KEY=<your_stripe_key>  # Optional
REDIS_URL=redis://localhost:6379     # Optional

# Network
X402_DEFAULT_NETWORK=eip155:84532    # Base Sepolia dev
CDP_API_KEY_ID=<cdp_id>              # For mainnet selling
CDP_API_KEY_SECRET=<cdp_secret>

🔧 开发

运行测试

# Full test suite
make test

# Specific tests
pytest tests/test_mcp_tools.py -v
pytest tests/test_commerce.py -v

# Dashboard tests
cd dashboard && pnpm vitest run

添加新的 MCP 工具

按照 app/tools_registry.py 中的模式进行操作:

  1. 将工具定义添加到 TOOL_SPECS 列表

  2. 在 app/intelligence/ 目录中实现相应的处理程序

  3. 在 app/mcp_server.py 中注册

  4. 在 README 中新增工具表格行

  5. 运行 scripts/capture_goal_evidence.py 更新已记录的测试结果

要记住: 工具注册表是唯一的事实来源!必须同步修改所有相关文件,否则 CI 将会失败。

CI/CD 流程

GitHub Actions 包括:

  • ✅ Pytest(Python 3.12)

  • ✅ Vitest(仪表盘)

  • ✅ 安全扫描(秘密扫描、SAST)

  • ✅ 覆盖率检查(>80%)

  • ✅ Docker 构建与推送

📊 可观测性

指标端点

  • /stats - 实时使用统计数据

  • /ledger/{name} - 结算历史记录

  • /quota/{agent_id} - 剩余配额

  • /swarm/stats - 自治智能体指标

事件流

订阅 SSE 服务 /events 以获取实时:

  • 工具执行日志

  • 结算确认

  • 配额更新

  • Swarm 阶段转换

🔒 安全性

  • ✅ 不会提交任何私钥到仓库(.env 始终被忽略)

  • ✅ 对所有 URL 探测启用 SSRF 防御

  • ✅ 按 agent_id 对请求频率进行限制

  • ✅ 结算使用幂等键

  • ✅ 支持通过 KEY_PROVIDER 服务接口进行密钥轮换

  • ✅ 绝不会在响应中暴露异常内部实现信息

卖家与买家密钥:

  • 收款地址(接收地址): 冷钱包用于接收收益(与买家密钥分开!)

  • 买家密钥: 仅用于支付上游成本的热钱包(绝不用于接收)

  • 绝对不要在 Render 上设置 EVM_PRIVATE_KEY - 仅本地使用!

|## 🌐 部署

在线仪表盘: https://alpha-sentinel-mcp.vercel.app/ (即将推出)

快速部署到 Vercel

点击下方链接或按照 DEPLOYMENT.md 中的步骤操作:

Deploy with Vercel

手动部署

# Install Vercel CLI
npm install -g vercel

# Deploy dashboard
cd /home/keef/alpha-sentinel-mcp
vercel --prod

仪表盘将自动代理 API 调用到您的个人后端。

Render(API 服务)

通过 render.yaml 部署 - 自动配置:

  • Web 服务(端口 8403)

  • Redis 实例(自动配置)

  • 环境变量来自安全存储

  • 健康检查端点:/health

Docker

docker build -t alpha-sentinel-mcp .
docker run -p 8403:8403 \
  -e X402_PAY_TO_ADDRESS=$PAY_TO \
  -e EVM_PRIVATE_KEY=$BUYER_KEY \
  alpha-sentinel-mcp

🤝 Swarm 智能体集成

当 SWARM_ENABLED=true 时,Alpha Sentinel 可以:

  1. 侦察(Scout) 探索上游 x402 数据源

  2. 守卫(Warden) 强制实施每个 ledger/policy.json 的支出限额

  3. 财务(Treasurer) 支付上游数据费用(唯一的支出角色)

  4. 档案(Archivist) 组装付费集合产品

  5. 主权(Sovereign) 重新定价以达到目标 LTV:CAC ≥ 3.0

  6. 商户(Merchant) 列出生成的合成产品

集合示例:购买来自 3 个上游的原始价格 → 聚合 → 以 3 倍加价重新销售为“高级套装”!

📈 市场发现

在 CDP Bazaar 上展示:

  • 服务名称:Alpha Sentinel 市场情报平台

  • 标签:基础、智能、加密、市场、情绪

  • 所有权证明:使用接收地址签名

  • 产品列表 URL:https://alpha-sentinel-mcp.onrender.com/.well-known/x402

🎓 文档

🏆 成功指标

  • 警报延迟: 端到端 <30s

  • 结算成功率: >95% 首次尝试

  • LTV:CAC 比率: 目标 ≥ 3.0

  • 锁仓边际: 组合产品 ≥ 50%

  • 配额利用率: 超过 70% 的专业层级转化率

🎯 设立

由 Keith 构建 (kwizzlesurp10-ctrl)作为 AI Agency Factory 生态的一部分。灵感来源于 Karpathy 的 LLM Wiki 知识组织方式和 x402 微支付革命!

特别感谢:

  • Coinbase CDP 团队提供 x402 促进层

  • FastAPI + FastMCP 维护者们

  • 整个开源 x402 社区

📄 开源许可证

MIT - 允许商业使用,请注明来源! ★彡


准备好实现你的市场情报价值了吗?
👉 立即部署
👉 阅读用户指南
👉 查看仪表盘

(◕‿◕) ✨ 预测智能,由链上经济学提供动力! ♪

Available Tools

13 tools
aggregate_sentimentA
Read-onlyIdempotent
Inspect

Aggregate social media sentiment across multiple sources.

Combines Twitter/X, Reddit, and other platforms into unified sentiment scores (-1.0 to +1.0). Detects momentum shifts and trend changes in community opinion.

Use Case: Gauge market psychology before major moves.

Cost: $0.01 USD per call

Sources: twitter, reddit, telegram, news

Output: Overall sentiment category (bullish/neutral/bearish) with breakdown by source.

Inputs: symbols (required), sources, window_minutes.

analyze_volatilityA
Read-onlyIdempotent
Inspect

Analyze price volatility using statistical Z-score anomaly detection.

Calculates standard deviations from mean price to detect anomalous movements. Configurable thresholds and time windows for customized analysis.

Use Case: Identify unusual price movements before they become obvious trends.

Cost: $0.02 USD per call

Parameters:

  • symbol: Crypto symbol to analyze

  • window_minutes: Historical window (1-1440 min, default: 60)

  • z_threshold: Std dev threshold for anomalies (0.1-5.0σ, default: 2.0)

Alert Levels: NORMAL → MODERATE → HIGH → CRITICAL

Inputs: symbol (required), z_threshold, window_minutes.

calculate_risk_scoreA
Read-onlyIdempotent
Inspect

Calculate multi-factor risk assessment for cryptocurrencies.

Combines volatility, liquidity, market cap tier, and Bitcoin correlation into a comprehensive risk score (0-1 scale) with actionable recommendations.

Use Case: Assess position sizing and stop-loss levels.

Cost: $0.03 USD per call

Risk Levels: LOW | MEDIUM | HIGH | CRITICAL

Factors Analyzed:

  • Volatility risk (40% weight)

  • Liquidity risk (25% weight)

  • Market cap tier (20% weight)

  • BTC correlation (15% weight)

Inputs: symbols (required), include_factors.

fetch_priceA
Read-onlyIdempotent
Inspect

Fetch real-time cryptocurrency price from CoinGecko API.

Returns current price, 24h change, and trading volume for any supported cryptocurrency.

Use Case: Get instant price data before executing trades or analyzing positions.

Cost: $0.005 USD per call (cheapest tier - high frequency allowed)

Example Response: { "symbol": "bitcoin", "price_usd": 67842.50, "change_24h": 2.34, "volume_24h": 28500000000, "timestamp": "2026-08-24T15:30:00Z" }

Inputs: symbol (required).

generate_market_reportA
Read-onlyIdempotent
Inspect

Generate comprehensive market intelligence report.

Synthesizes price, volatility, sentiment, and risk data into actionable insights. Perfect for daily briefings or strategic planning.

Use Case: End-of-day summary or pre-trade due diligence.

Cost: $0.15 USD per report

Report Types:

  • daily: 24-hour overview

  • weekly: 7-day trends

  • monthly: 30-day strategic analysis

Includes: Executive summary, key signals, actionable recommendations

Inputs: format, symbols, report_type.

tokenized_stock_corporate_actionA
Read-onlyIdempotent
Inspect

Fetch one announced/applied corporate action by UUID (from onchain Announcement URI).

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.02 USD per call

Inputs: corporate_action_id (required).

tokenized_stock_listB
Read-onlyIdempotent
Inspect

List Coinbase tokenized stocks on Base (read-only catalog).

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.01 USD per call

Inputs: query, page_size, page_token.

tokenized_stock_lookupA
Read-onlyIdempotent
Inspect

Lookup metadata for one tokenized stock (contract, ISIN, multiplier, prospectus).

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.01 USD per call

Inputs: symbol (required).

tokenized_stock_model_portfolioA
Read-onlyIdempotent
Inspect

Model portfolio / rebalance suggestions (weights only — no orders).

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.08 USD per call

Inputs: symbols (required), objective, current_weights.

tokenized_stock_nav_deviationB
Read-onlyIdempotent
Inspect

Compare Chainlink NAV/reference to underlying equity spot (research spread).

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.02 USD per call

Inputs: symbol (required).

tokenized_stock_peg_gap_radarA
Read-onlyIdempotent
Inspect

Peg-Gap Radar: onchain DEX price vs underlying reference, alerts over threshold %.

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.03 USD per call

Inputs: symbols, page_size, alert_threshold_pct.

tokenized_stock_research_briefB
Read-onlyIdempotent
Inspect

Multi-symbol research brief: metadata, NAV staleness, deviation notes.

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.05 USD per call

Inputs: symbols (required).

tokenized_stock_seal_receiptB
Idempotent
Inspect

Seal a Proof-of-Research receipt hash on Base (ResearchReceiptRegistry).

Prepares unsigned calldata by default; optional Sepolia submit via env. Not trading.

RESEARCH ONLY — not investment advice. No trading, execution, custody, or swaps. Coinbase tokenized stocks are not offered to U.S. persons.

Cost: $0.03 USD per call

Inputs: content_hash (required), brief_id.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 13 tools

    Disambiguation3/5

    The set splits into two clusters that are individually readable, but within the tokenized_stock cluster several tools measure the same thing: nav_deviation and peg_gap_radar both compare onchain/DEX price to an underlying reference, and research_brief explicitly bundles metadata + NAV staleness + deviation, substantially overlapping lookup, corporate_action, and nav_deviation. The general crypto tools (fetch_price, analyze_volatility, aggregate_sentiment, calculate_risk_score, generate_market_report) are more distinct, though generate_market_report synthesizes the others and an agent could plausibly pick it instead of the individual signals.

    Naming Consistency3/5

    Five tools follow a clean verb_noun convention (fetch_price, analyze_volatility, aggregate_sentiment, calculate_risk_score, generate_market_report), while the other eight switch to a noun namespace prefix (tokenized_stock_list, tokenized_stock_lookup, tokenized_stock_seal_receipt). Each convention is internally consistent and the namespace is readable, but mixing verb-first and noun-prefixed forms in one server is a notable deviation from a single predictable pattern.

    Tool Count4/5

    Thirteen tools sits comfortably in the well-scoped range, and each maps to a concrete operation with a stated cost tier. The count is slightly heavy because roughly four of the tokenized-stock tools (lookup, nav_deviation, research_brief, peg_gap_radar) could likely collapse into two without losing capability.

    Completeness3/5

    The tokenized-stock surface is broad (catalog, metadata, corporate actions, NAV, peg, portfolio, receipt), but tokenized_stock_corporate_action only fetches by UUID with no announcement-listing tool to discover those UUIDs, which is a real dead end. On the general crypto side, price/volatility/sentiment/risk/report are covered but there is no historical OHLC or liquidity/order-book data to feed the volatility and risk models.

    Maintenance

    ActivityActive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Provides AI agents with real-time, structured data feeds including news, trends, market sentiment, and crypto prices via MCP tools.
      MIT
    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables MCP-compatible AI clients to access live crypto market data and AI-driven quantitative analysis, with structured outputs and full observability.
      -
    • A
      license
      A
      quality
      B
      maintenance
      Live Base mainnet (eip155:8453) x402 micropayment MCP gateway by Keith S. (kwizzlesurp.com · Sevtech · Dragon&PandaDuo). Pairs a free open-data catalog with $0.01 USDC property & rental compliance checks across 14+ US cities, Base gas/finality intelligence, and x402 buyer/seller tools. Verified on x402-trust (82.5/100, 100% uptime) & 24K Labs.
      5
      20
      1
      MIT