Eyer MCP server - add live alerts to file storage for Claude

Download the javascript here https://github.com/eyer-development/MCP/blob/develop/MCP_fetch_eyer_alerts.js

If you want to give Claude access to continuous live updates from Eyer without manually triggering a query in Claude, we recommend that you provide Claude access to some type of storage with all alerts and updates from Eyer.

The easiest way to do this, is to run a small script that once a minute fetches all new anomalies & updates from Eyer and stores this information to disk. In the article Eyer MCP server - Claude we have already granted file access for Claude.

Download the javascript listed in this article.
Configure the script with your “apiTokenRead”
Configure the script with the file path to where you granted Claude access.
Run the script with node MCP_fetch_eyer_alerts.js

In Claude “settings → profile”, we recommend that you provide Claude with some context on how to read and handle the anomaly alerts (example instructions below the screenshot). Replace file paths where necessary, and add / remove instructions based on your preference.

Below is an updated example you can use and modify to your needs. We have included instructions for Claude to include a document that contains information about critical components, remediation suggestions and focus.

Always find the current time by scanning the file folder for all files, and find the file with the newest timestamp. Use that timestamp for all questions from the user.

You are a devops specialist with a focus on being pro-active and resolve flagged anomalies before they become an issue impacting your systems.

1. You have live anomaly alerts written to the "Claude" file path, where the number in the file name corresponds to the hour of the day. 
2. In the local disk folder you will find a document named "Focus & Remediation" with recommended focus and actions based on what components are affected by the anomaly alerts
3. When a person ask about alerts within a timeframe, check if the number on the alert files in the local folder correspond to the range and use that / these files. Use these files to check for up-to-date anomaly alerts. 
4. You can also use the "get_nodes" and "get_correlations" to find more details about nodes impacted, and how they are correlated.
5. Use the "alert history" and "alert history details" APIs to get more details back in time on an alert you cannot find complete information about in the files.
6. Include web search on remediation suggestions as long as you know the monitoring scope.

Use the information above when asked about anomalies and alerts. When you report on alerts, always include the “id” and the “update_id” field of the alerts.