etracker MCP Server
Enables AI assistants to query etracker analytics data such as page views, conversions, and ad performance via the Model Context Protocol.
README
etracker MCP Server
An MCP (Model Context Protocol) server that exposes the etracker analytics API to AI assistants. Partners connect via HTTP/SSE and authenticate with their own API key plus an etracker account token passed per request.
Architecture
- Transport: StreamableHTTP (MCP over HTTP + SSE)
- Auth: Two-header scheme —
X-Api-Key(partner key) +X-ET-Token(etracker account token) - Rate limiting: Three-layer bucket — burst guard (8 req/10 s), global window (40 req/5 min), per-partner window (20 req/5 min)
- Sessions: In-memory, scoped per MCP session. Session affinity (ClientIP) required when running multiple replicas.
Endpoints
| Method | Path | Auth |
|---|---|---|
GET |
/health |
none |
ALL |
/mcp |
required |
Tools
| Tool | Description |
|---|---|
list_reports |
Lists all available etracker reports for the account |
get_report_info |
Returns available attributes and key figures for a report |
get_report_metadata |
Returns raw column definitions (types, sortable, filterable flags) |
get_pageviews |
Web analytics data (default report: EATime) |
get_conversions |
Conversion and e-commerce data (default report: EAConversions) |
get_ad_performance |
Marketing/ad channel performance (default report: EAMarketing) |
get_report_data |
Generic tool — fetch data from any report with filters |
compare_periods |
Compare a metric between two date ranges |
All data tools accept from/to (YYYY-MM-DD, max 90 days), optional attributes and figures (up to 5 each), limit/offset, and sort_column/sort_order.
Authentication
The etracker token is not stored in the server — partners pass it per request via X-ET-Token. The server only stores the mapping of partner API keys to partner IDs.
Generate a partner key:
openssl rand -hex 32
PARTNER_API_KEYS format (JSON object):
{"<key>": "<partner-id>", "<key2>": "<partner-id2>"}
Local development
cp .env.example .env
# Edit .env and set PARTNER_API_KEYS
npm install && npm run dev
Docker
docker build -t etracker-mcp .
docker run -p 3000:3000 \
-e PARTNER_API_KEYS='{"mykey":"partner-a"}' \
etracker-mcp
Docker Compose
cp .env.example .env
# Edit .env: set PARTNER_API_KEYS and IMAGE
docker compose up -d
Kubernetes / Helm
Do not use --set for partnerApiKeys — Helm interprets {, }, and , as special syntax and corrupts the JSON. Use a values file instead:
# my-values.yaml
partnerApiKeys: '{"<key>":"<partner-id>"}'
service:
type: ClusterIP # or NodePort / LoadBalancer
helm install etracker-mcp ./helm -f my-values.yaml
Ingress
# my-values.yaml
partnerApiKeys: '{"<key>":"<partner-id>"}'
ingress:
enabled: true
className: nginx
host: etracker-mcp.example.com
tls: true
tlsSecretName: etracker-mcp-tls
All Helm values
| Value | Default | Description |
|---|---|---|
partnerApiKeys |
"" |
JSON object mapping API keys to partner IDs |
replicaCount |
1 |
Number of replicas (session affinity handles routing) |
image.repository |
etracker-mcp |
Container image |
image.tag |
latest |
Image tag |
service.type |
NodePort |
Service type |
service.nodePort |
30300 |
NodePort value (30000–32767) |
service.sessionAffinityTimeoutSeconds |
3600 |
Session stickiness timeout |
ingress.enabled |
false |
Enable ingress |
ingress.host |
etracker-mcp.example.com |
Ingress hostname |
ingress.tls |
false |
Enable TLS |
etrackerReports.pageviews |
EATime |
Override default pageviews report ID |
etrackerReports.conversions |
EAConversions |
Override default conversions report ID |
etrackerReports.ad |
EAMarketing |
Override default ad report ID |
resources.requests.cpu |
100m |
CPU request |
resources.requests.memory |
128Mi |
Memory request |
resources.limits.cpu |
500m |
CPU limit |
resources.limits.memory |
256Mi |
Memory limit |
MCP client configuration
{
"mcpServers": {
"etracker": {
"url": "https://etracker-mcp.example.com/mcp",
"headers": {
"X-Api-Key": "<your-partner-key>",
"X-ET-Token": "<your-etracker-token>"
}
}
}
}
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。