monitor_skill

Overview

This skill automates periodic webset updates by creating and managing monitors that either search for new items or refresh existing items. It provides command patterns, output-format guidance, and operational rules to keep websets fresh and within API limits. Use it to schedule reliable, repeatable collection updates with clear separation of responsibilities.

How this skill works

The skill issues exa-ai monitor commands to create monitors tied to a webset ID with cron schedules and behavior types (search or refresh). It recommends running commands directly and parsing JSON output in follow-up steps, avoiding nested shell substitutions. It also enforces schema and output-format rules so downstream parsing and automation remain predictable.

When to use it

• When you need to add new items to a webset on a schedule (search monitors).
• When you need to refresh metadata or content for existing items periodically (refresh monitors).
• When you want robust, repeatable automation without fragile nested shell commands.
• When you need to manage rate limits by scheduling heavy operations during off-peak hours.
• When you want consistent output formats for programmatic parsing with jq or schemas.

Best practices

• Create separate monitors for search and refresh to keep responsibilities distinct.
• Schedule refresh monitors during off-peak/nighttime to reduce rate-limit impact.
• Run commands directly and parse their JSON output in a separate step; avoid nested $(...) substitutions.
• Choose one output approach (toon for reading, JSON + jq for parsing, or schemas + jq for structured extraction).
• Start with conservative schedules (daily/weekly) and increase frequency only after monitoring execution history.

Example use cases

• Daily morning search monitor to append new items discovered for competitive intelligence.
• Nightly refresh monitor to update scores, metadata, or content freshness for a large collection.
• Weekly summary runs that use schemas + jq to produce consistent structured reports for downstream pipelines.
• Create a webset, capture its webset_id as JSON, then create paired search and refresh monitors with clear cron and timezone settings.
• Inspect monitor runs and execution history to debug failures or tune cron schedules.