pulsemon_skill

Overview

This skill integrates with PulseMon to monitor cron jobs and background tasks. It lets you list monitors, inspect details and incidents, create or modify monitors, and manage alerts and pause/resume state. The skill favors clear, human-readable summaries of status and durations for quick incident triage.

How this skill works

The skill calls PulseMon API endpoints under https://pulsemon.dev/api/v1 using a Bearer token from the PULSEMON_API_KEY environment variable. It lists monitors (name, status, last ping), fetches monitor details (recent pings, incidents), and performs create/update/delete/pause/resume actions. Responses are normalized and durations converted into readable formats (e.g., "2 hours ago").

When to use it

• Check overall health of scheduled jobs and background processes.
• Create a new monitor for a job you just deployed or scheduled.
• Investigate recent failures or incidents for a specific monitor.
• Temporarily pause monitoring during maintenance windows.
• Adjust expected intervals or grace periods after changing job frequency.

Best practices

• Provide a descriptive name and URL-safe slug (lowercase letters, numbers, hyphens) when creating monitors.
• Set expectedInterval using common presets (e.g., hourly = 3600, daily = 86400) and use sensible grace periods—default 10% of interval, minimum 30s—if unspecified.
• Confirm deletion before removing a monitor to prevent data loss and accidental downtime alerts.
• When listing monitors, prioritize showing name, status, and humanized last ping times for fast triage.
• Use tags to group monitors by environment or service for easier filtering.

Example use cases

• "Check my monitors" summarizes all monitors with status and how long since last ping, highlighting any that are down.
• Create a monitor for a nightly backup: suggests expectedInterval=86400 and a sensible gracePeriod, then returns the ping URL https://pulsemon.dev/api/ping/nightly-backup.
• Pause a monitor during deployments and resume after verifying pings to avoid false incidents.
• Fetch recent incidents for a failed background task to see when it went down and how many retries occurred.
• Update a monitor when you change a job schedule (e.g., every hour -> expectedInterval=3600) and adjust gracePeriod accordingly.