SitePulsar MCP Server
Measures and improves how findable, readable, and usable a website is to AI answer engines and autonomous AI agents through hosted AEO audit tools.
README
SitePulsar MCP Server
Agent Experience Optimization (AEO) for AI agents — measure and improve how findable, readable, and usable any website is to AI answer engines and autonomous AI agents, directly from your MCP client.
A remote, hosted Model Context Protocol server. There is nothing to build or run locally — connect your MCP client to the hosted endpoint and the 13 tools below become available.
Endpoint (streamable-http): https://mcp.sitepulsar.ai/mcp
What is SitePulsar / AEO?
SitePulsar is an Agent Experience Optimization (AEO) audit platform. It measures how findable, readable, and usable a website is to AI answer engines and autonomous AI agents, and returns one AEO score across three pillars:
- FIND — discoverability across ChatGPT, Claude, Gemini, Perplexity, and Grok.
- READ — comprehension, structured data, and trust signals.
- USE — agent-actionability (can an autonomous AI agent actually take action on the site?).
AEO is the agent-native successor to what used to be split as GEO + AIO. SitePulsar is agent infrastructure, not an SEO tool.
Quick start
The server supports two auth modes. Pick whichever fits your client.
Option A — OAuth (recommended for Claude; no API key)
Add SitePulsar as a custom connector in Claude using the endpoint. Claude walks you through the OAuth handshake; no API key needs to be stored.
- Endpoint:
https://mcp.sitepulsar.ai/mcp - OAuth protected-resource metadata:
https://mcp.sitepulsar.ai/.well-known/oauth-protected-resource
Option B — Bearer API key (for headless / non-interactive clients)
Mint an MCP API key in the SitePulsar dashboard (Pro/Business plan), then send it on every request:
Authorization: Bearer <mcp_api_key>
See examples/headless-curl.md for raw HTTP usage.
Connect matrix
| Client | How to connect |
|---|---|
| Claude (desktop / web) | Settings → Connectors → Add custom connector → paste https://mcp.sitepulsar.ai/mcp, then complete OAuth. |
| Claude Code (CLI) | claude mcp add --transport http sitepulsar https://mcp.sitepulsar.ai/mcp |
| Cursor / Cline (JSON config) | Add to your MCP config: see below. |
Cursor / Cline JSON config:
{
"mcpServers": {
"sitepulsar": {
"url": "https://mcp.sitepulsar.ai/mcp"
}
}
}
A ready-to-paste config is in examples/claude_desktop_config.json.
For headless clients that cannot do interactive OAuth, add an
Authorization: Bearer <mcp_api_key>header (see Option B above).
Tools
13 tools, grouped by pillar coverage. "Metered" tools run a billed audit/scan and require a Pro/Business plan; "Free" tools are read-only or lightweight probes.
| Tool | What it does | Cost |
|---|---|---|
check_agent_readiness |
Fast synchronous agent-readiness check for a URL (no full run, no charge). Immediate signals on machine readability, structured data, and agent-accessibility. | Free |
run_audit |
Start a full async AEO audit for a URL. Returns an audit_id; poll with get_audit. |
Metered (Pro/Business) |
get_audit |
Poll status and, when complete, a decision-ready summary: AEO score (overall + FIND/READ/USE), weakest pillar, headline takeaways, and top fixes. | Free |
get_audit_detail |
Full per-section breakdown of a completed audit — per-dimension sub-scores, reputation across engines, competitor cluster, crawl/schema/robots, agentic + USE probes, rendered-DOM. | Free |
get_audit_full |
One call returning summary + full detail + fixes (use expand to trim). |
Free |
compare_aeo |
Compare AEO posture across multiple URLs on the same pillar scale. | Free |
get_fixes |
Prioritised, pillar-tagged action plan for a completed audit, ordered by estimated impact. | Free |
search_companies |
Samples major AI engines for which companies they name for a query; returns a consensus shortlist (≤5). | Free |
probe_agent_discovery |
Checks the official MCP registry, PyPI, and GitHub for servers/packages tied to a domain or brand. | Free |
probe_mcp_functional |
Discovers a site's advertised MCP endpoint and reports its declared OAuth/transport posture + declared tool names. | Free |
probe_ucp_readiness |
Inspects /.well-known/ucp for AI-shopping-agent transactability (presence + advertised capabilities). |
Free |
scan_visibility |
Live AI-visibility scan for any brand (crawl + reputation across engines): where a brand is mentioned. | Metered (Pro/Business) |
scan_product_page |
Product-page readability scan for AI agents; structured product/offer extraction. | Free or metered — see dashboard |
Full machine-readable tool reference: llms.txt.
Authentication
SitePulsar accepts either of the following on the hosted endpoint.
OAuth
Used by interactive clients (e.g. Claude custom connectors). The server advertises its OAuth posture via RFC 9728 protected-resource metadata:
https://mcp.sitepulsar.ai/.well-known/oauth-protected-resource
Your client discovers the authorization server from that document and completes the handshake — no API key is stored on your side.
Bearer API key
Used by headless clients. Mint a key in the SitePulsar dashboard (Pro/Business plan) and send it on every request:
Authorization: Bearer <mcp_api_key>
Metered tools (run_audit, scan_visibility, and the metered path of scan_product_page) require a Pro/Business plan regardless of auth mode.
Links
- Website: https://www.sitepulsar.ai
- Tool reference (llms.txt): https://www.sitepulsar.ai/llms.txt
- Hosted MCP endpoint:
https://mcp.sitepulsar.ai/mcp - OAuth metadata:
https://mcp.sitepulsar.ai/.well-known/oauth-protected-resource - Listed in the official MCP registry as
ai.sitepulsar/mcp(v1.0.0).
License
MIT — see LICENSE.
Built by
SitePulsar — EU-based (Netherlands). Founder & CEO: Gijs van Bueren. Questions: info@sitepulsar.ai.
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。