kaito-mcp-server
Provides crypto market intelligence tools, reference resources, and prompt templates for sentiment analysis, mindshare tracking, social intelligence, and more.
README
kaito-mcp-server
MCP proxy for Kaito AI crypto market intelligence API. Provides crypto market intelligence tools, reference resources, and prompt templates for sentiment analysis, mindshare tracking, social intelligence, and more.
Getting Started
A Kaito API key is required. Get one from Telegram or email support@kaito.ai.
Option A: Direct HTTP
For MCP clients that support remote HTTP servers and let you set a static Authorization header (e.g., Claude Code):
{
"mcpServers": {
"kaito": {
"type": "http",
"url": "https://bff.kaito.ai/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
No local installation required.
Note: Clients whose remote-MCP UI is OAuth-only — including Claude Desktop's Custom Connectors and Claude.ai (web) — are not compatible with Kaito's static Bearer-token auth. Use Option B on those clients.
Option B: npx (stdio proxy)
For MCP clients that only support local stdio servers:
{
"mcpServers": {
"kaito": {
"command": "npx",
"args": ["-y", "kaito-mcp-server@latest"],
"env": {
"KAITO_API_KEY": "your-api-key"
}
}
}
}
This runs a thin local proxy that forwards all requests to the Kaito API over HTTP.
Client-specific setup
Claude.ai (web)
Not currently supported — Claude.ai web's Custom Connectors UI requires OAuth, but the Kaito API uses static Bearer tokens. Use Claude Code or Claude Desktop with Option B instead.
Claude Desktop (one-click)
Download the latest kaito-mcp-server.mcpb and open it — Claude Desktop will handle the rest.
Note: MCPB installs a pinned version and does not auto-update. To get the latest version, re-download the
.mcpbfrom the latest release — or switch to the standard configuration which uses@latestto always pull the newest version.
Claude Desktop (manual)
Add Option B to your claude_desktop_config.json:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Note: Option A (HTTP) is not supported here —
claude_desktop_config.jsononly accepts stdio entries, and Claude Desktop's Settings → Connectors UI requires OAuth.
Claude Code
HTTP (recommended):
# If you previously used the npx setup, remove it first:
claude mcp remove kaito
# Then add the HTTP server:
claude mcp add kaito --transport http https://bff.kaito.ai/api/mcp --header "Authorization: Bearer YOUR_API_KEY"
stdio (alternative):
claude mcp add kaito -e KAITO_API_KEY=your-api-key -- npx -y kaito-mcp-server@latest
VS Code
Add the following to .vscode/mcp.json (workspace) or your user-profile mcp.json (open via the MCP: Open User Configuration command).
HTTP (recommended):
{
"servers": {
"kaito": {
"type": "http",
"url": "https://bff.kaito.ai/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
stdio (alternative):
{
"servers": {
"kaito": {
"command": "npx",
"args": ["-y", "kaito-mcp-server@latest"],
"env": {
"KAITO_API_KEY": "your-api-key"
}
}
}
}
Cursor
Add to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global).
HTTP (recommended):
{
"mcpServers": {
"kaito": {
"url": "https://bff.kaito.ai/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
stdio (alternative): paste Option B.
Other MCP Clients
For any MCP-compatible client, the server can be started with:
KAITO_API_KEY=your-api-key npx -y kaito-mcp-server@latest
The transport is stdio. Use this command in your client's MCP server configuration.
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
KAITO_API_KEY |
Yes | — | Your Kaito API key |
KAITO_MCP_URL |
No | https://bff.kaito.ai/api/mcp |
Override the remote MCP endpoint (e.g., for staging) |
Recommended Skills
Pair this MCP server with kaito-skills — curated Agent Skills (social listening, mindshare pulse, etc.) that use the tools below. Works in Claude Code, Cursor, VS Code + Copilot, and 40+ other agents.
npx skills add MetaSearch-IO/kaito-skills
See the kaito-skills README for client-specific install commands.
Tools
Reference
| Tool | Description |
|---|---|
kaito_entities |
List or search supported Kaito token identifiers by project name, token value, symbol, or CoinGecko slug |
kaito_narratives |
List or search supported Kaito narrative IDs by narrative code or display name |
Search & Discovery
| Tool | Description |
|---|---|
kaito_search |
Default natural-language Kaito search for most user requests across Twitter, News, Research, and Podcast content |
kaito_feeds |
Use this when you want the top ranked content feed for the whole market or scoped to a single resolved token, optionally within a time window. |
kaito_advanced_search |
Power-user structured Kaito search with explicit control over tokens, usernames, sources, time filters, sorting, language, and thresholds |
Sentiment, Mindshare & Engagement
| Tool | Description |
|---|---|
kaito_tweet_engagement_info |
Detailed engagement metrics for a specific tweet, including likes, retweets, replies, views, and smart engagement count |
kaito_sentiment_entity |
Daily sentiment time series for a token, including volume-weighted bullish and bearish scores plus notable events |
kaito_engagement |
Daily engagement metrics for a token or keyword, including total and smart/KOL engagement |
kaito_mentions |
Daily mention counts for a token or keyword across Twitter, News, Podcast, and Research |
kaito_mindshare_entity |
Daily mindshare time series for a token, showing its share of crypto Twitter conversation |
kaito_mindshare_entity_arena |
Project rankings by mindshare score, with optional category and time window filters |
kaito_mindshare_entity_delta |
Top gainers and losers by mindshare change over a selected time window |
kaito_mindshare_narrative |
Daily mindshare time series for a crypto narrative |
Account & KOL Analytics
| Tool | Description |
|---|---|
kaito_smart_following_market |
Market-level signal showing which accounts smart followers have recently followed |
kaito_smart_followers |
Smart follower data for a Twitter user, including total smart follower count or gained followers for a specific date |
kaito_smart_following |
Latest smart accounts followed by a Twitter user in reverse chronological order |
kaito_mindshare_entity_by_account |
Top KOLs ranked by mindshare for a given token |
kaito_twitter_user_metadata |
Twitter user profile metadata by user ID, including name, username, bio, follower stats, and account classification |
Events & Catalysts
| Tool | Description |
|---|---|
kaito_events |
Upcoming catalyst events for a token, with filtering by event type, source, and date range |
Resources
| Resource | URI | Description |
|---|---|---|
| Tokens | kaito://tokens |
All supported token values, symbols, and project names (no auth required) |
| Narratives | kaito://narratives |
All supported narrative IDs (no auth required) |
Prompt Templates
| Prompt | Description |
|---|---|
analyze_token |
Deep-dive on a single token: top discussions via search, sentiment, mindshare trends, and upcoming events |
market_roundup |
Priority-ranked briefing of top stories with smart account signals, mindshare movers, and catalysts |
discover_trending |
Trending discovery: mindshare arena rankings, delta movers, smart account signals, narrative shifts, and deep-dive search per top gainer |
social_listening |
Full social analysis: key discussions, sentiment, mentions, engagement, KOL mindshare, smart followers, and optional competitor comparison |
watchlist_portfolio |
Monitor developments, upcoming catalysts, and key narratives across a set of tokens with mindshare, sentiment, and events |
Development
git clone https://github.com/MetaSearch-IO/kaito-mcp-server.git
cd kaito-mcp-server
npm install
npm run build
Testing
Integration tests hit the real Kaito API via the proxy:
KAITO_API_KEY=your-key npm test
Tests are skipped if KAITO_API_KEY is not set.
MCP Inspector
KAITO_API_KEY=your-key npx @modelcontextprotocol/inspector node build/index.js
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。