tabctl_skill

Overview

This skill lets you manage and analyze Microsoft Edge tabs and groups using tabctl. It emphasizes read-only inspection and reporting by default, and performs mutating actions only when explicit targets and confirmations are provided. Use it to discover stale or duplicate tabs, produce reports, and safely organize tabs and groups.

How this skill works

tabctl queries the browser to list windows, groups, and tabs, then runs analyses (stale detection, duplicates, metadata extraction) and can produce JSON or markdown reports. It supports targeted inspections (page metadata, selectors, screenshots) with wait modes to handle dynamic pages. Mutating actions require explicit scope flags (--tab, --group, --window) and confirmation to avoid accidental changes.

When to use it

• Audit open tabs to find stale, duplicate, or unused pages
• Generate markdown or JSON reports for tab hygiene or handoff
• Search or filter tabs by URL, title, or selector results
• Inspect a tab’s page metadata or extract links safely
• Organize or close specific tabs or groups after confirming targets

Best practices

• Prefer read-only commands: list, analyze, inspect, report
• Scope mutating commands with --tab, --group, or --window and use --confirm
• Avoid mass-mutating commands (never run archive --all or close --apply in normal sessions)
• Use --json for scriptable output and pipe to jq/node for custom filtering
• Use --wait-for settle for JS-heavy pages to get stable titles and metadata

Example use cases

• Find tabs not used in 7+ days: tabctl analyze --stale-days 7 --json
• Extract all absolute links from a page: tabctl inspect --tab <id> --selector '{"name":"links","selector":"a[href]","attr":"href-url","all":true}'
• Create a markdown report of a window: tabctl report --format md --window active
• Refresh a single tab after fixing a site: tabctl refresh --tab <id>
• Take a full-page screenshot for visual context: tabctl screenshot --tab <id> --mode full