servicenow-agent_skill

Overview

This skill provides read-only CLI access to ServiceNow Table, Attachment, Aggregate, and Service Catalog APIs. It is designed for safe inspection: schema discovery, record reads, attachment download, aggregates, and ticket history retrieval without any create/update/delete capability. Use environment variables or command flags to authenticate and keep queries small to avoid large payloads.

How this skill works

The CLI issues only GET requests against the ServiceNow Table, Attachment, Aggregate, and Service Catalog OpenAPI endpoints. It reads auth from a .env file by default (SERVICENOW_DOMAIN, SERVICENOW_USERNAME, SERVICENOW_PASSWORD) and supports overriding via flags. Commands include list, get, attach, stats, schema, history, sc, and batch to run multiple read presets.

When to use it

• Inspect table schemas before building integrations or queries.
• Retrieve individual records or small lists for diagnostics and audits.
• Download attachment metadata and binary files for local analysis.
• Run aggregates and counts without exporting large datasets.
• Read ticket conversation history and journal notes for troubleshooting.

Best practices

• Always use --sysparm_limit and --sysparm_fields to keep result sizes small.
• Run schema table first to discover exact field names and types.
• Prefer stats for counts and aggregates instead of fetching many rows.
• Use --sysparm_display_value=false when you need raw sys_id values.
• Use batch presets for repeatable, conservative read workflows.

Example use cases

• Run node cli.mjs schema incident to list valid incident fields before querying.
• List recent incidents: node cli.mjs list incident --sysparm_limit 5 --sysparm_fields number,short_description,priority,sys_id
• Fetch a ticket timeline: node cli.mjs history incident <sys_id> --pretty to read full comments and work notes.
• Download an attachment: node cli.mjs attach file <sys_id> --out /tmp/attachment.bin after locating metadata with attach list.
• Run aggregate counts: node cli.mjs stats incident --sysparm_query "active=true^priority=1" --sysparm_count true