
AyeWatch
io.github.SRX9v1.0.0更新於 Oct 7, 2026
Manage AyeWatch monitors for topics and web pages, and the webhook that receives their alerts.
概覽
讓助理管理 AyeWatch 針對主題與網頁的監控項目,並設定接收其警示的 webhook。
- 功能
- 將助理連線到託管的 AyeWatch 服務,使其能列出、查看、建立、更新、暫停、恢復與永久刪除監控主題或網頁網址的監控項目。它也能讀取與設定帳戶的 webhook 網址、開關投遞,以及輪換 webhook 簽章密鑰。警示本身不會透過 MCP 回傳,而是經由 AyeWatch 通知或你自己執行的 webhook 送達。
- 適用情境
- 當你希望助理在不開啟 AyeWatch 主控台的情況下,為某些主題與網頁建立或整理變更監控,或把警示導向自己的端點時,相當有用。若你主要想讀取警示歷史,則幫助不大,因為這些工具不會回傳該資料。
- 執行需求
- 位於 ayewatch.ai 的遠端 Streamable HTTP 端點,需要網路連線。需要付費方案的 AyeWatch 帳戶,API 與 MCP 存取已包含在內。需要 API 金鑰,以 Authorization 標頭依 Bearer aw_live_... 形式送出,可在 API 存取設定中建立。啟用的監控項目會消耗方案額度,預設每個金鑰每分鐘限 120 次請求。
安裝
在 SourceWeft 中
- 開啟 儀表板中的 AyeWatch,將其新增到工作區。
- 為需要使用其工具的對話啟用該服務。
Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。
其他 MCP 客戶端
把它新增到你客戶端的 mcpServers 設定中。
{
"mcpServers": {
"ayewatch": {
"type": "http",
"url": "https://ayewatch.ai/api/mcp"
}
}
}README
AyeWatch MCP
Connect an AI assistant to AyeWatch, which monitors topics and web pages and alerts you when something changes or matches what you care about. This repository holds the connection details and listing files for the AyeWatch MCP server. The server itself is hosted by AyeWatch.
- Endpoint:
https://ayewatch.ai/api/mcp - Transport: Streamable HTTP, stateless, no session ID
- Full documentation: https://ayewatch.ai/documentation/mcp
What you can do
The tools manage monitors and the webhook in your own account. They do not return alert history. The webhook tools are available only when you connect with an API key, not through OAuth, because the webhook receives every alert and its secret signs them.
Example prompts:
- "List my AyeWatch monitors and tell me which are paused."
- "Create a daily AyeWatch monitor for news about solid-state batteries."
- "Pause my AyeWatch monitor about the iPhone launch."
- "Send my AyeWatch alerts to https://example.com/hooks/ayewatch."
Alerts and webhooks
Alerts don't come back through MCP. They arrive through your AyeWatch notifications or a webhook. When a monitor detects new content, AyeWatch sends a POST with a JSON body (topic ID, headline, and the update) to an HTTPS endpoint that you run.
Set that endpoint with the set_webhook tool, the REST API (PUT https://ayewatch.ai/api/v1/webhook), or Notification settings. AyeWatch signs each delivery with a secret, sent in the X-AyeWatch-Signature header. The MCP tools never return that secret, so it stays out of AI conversations; read it in Notification settings or with GET https://ayewatch.ai/api/v1/webhook. AyeWatch also emails you whenever the webhook URL is set or changed. If you didn't make a change, revoke your API keys and rotate the secret. The payload fields, timeout, and retry behavior are in the webhooks documentation. The URL is yours, not AyeWatch's, so it is not part of the MCP connection files.
Requirements
- An AyeWatch account on a paid plan. API and MCP access are included with every paid plan.
- Active monitors use credits from your plan as checks run.
- Requests are limited to 120 per minute per API key by default.
Connect
Claude Code
Install the plugin from this repository. Claude Code asks for your API key when you enable the plugin and keeps it in your system's secure credential store.
Or add the server without the plugin:
Cursor
If you install the AyeWatch plugin from the Cursor Marketplace, set AYEWATCH_API_KEY when Cursor asks for the plugin's configuration. Otherwise, add this to your MCP configuration (~/.cursor/mcp.json):
VS Code
Add this to .vscode/mcp.json. VS Code asks for the key when the server starts and does not store it in the file.
Codex
Add this to ~/.codex/config.toml:
Any other MCP client
Use the endpoint above with the Streamable HTTP transport and send Authorization: Bearer aw_live_YOUR_API_KEY on every request.
Create an API key in API access settings. Keep it private and never commit it to a repository.
Sign in with OAuth
Registered apps can offer an AyeWatch sign-in instead of an API key. You review the access requested (read, or read and write) before choosing Allow access. Access expires after 30 days, or sooner if you disconnect. OAuth tokens work only on the MCP endpoint.
To end an app's access, open Connected apps and select Disconnect. Existing monitors keep their schedules, and copies the app already saved are not deleted.
What this repository contains
Nothing in this repository runs code on your machine. The plugin files only tell your client to connect to https://ayewatch.ai/api/mcp and send your API key to it in the Authorization header. AyeWatch receives the tool calls your assistant makes, such as a monitor's topic, URLs, and schedule, and nothing else from your machine.
Support and policies
- Support: https://ayewatch.ai/contact
- Privacy policy: https://ayewatch.ai/privacy
- Terms of service: https://ayewatch.ai/terms
License
MIT. The license covers the files in this repository, not the AyeWatch service.
來源:README.md,提交 fd13c1e
工具
0版本歷史
1- v1.0.0最新Oct 7, 2026

