x-bookmarks_skill

Overview

This skill fetches, summarizes, and manages your X/Twitter bookmarks so saved tweets become actionable work items instead of a graveyard of intentions. It auto-detects the best data source (bird CLI using browser cookies or X API v2 with OAuth tokens), categorizes bookmarks, and returns a digest that includes concrete actions the agent can take.

How this skill works

The skill first tries the bird CLI (fast, uses browser cookies) and falls back to the X API v2 if bird is unavailable or not authenticated. Bookmarks are fetched in a consistent JSON format, parsed for metadata (author, counts, media, quoted tweets), clustered by topic, and converted into a categorized digest that pairs each summary with recommended agent actions. It also supports state tracking for incremental digests and can run on a schedule via cron.

When to use it

• Ask for a quick or periodic digest: "check my bookmarks" or "bookmark digest"
• When you want new bookmarks summarized since your last review (scheduled or ad hoc)
• To search, categorize, or analyze saved tweets for content ideas or research
• When you need to clean up stale saves or unbookmark many items
• To recycle high-engagement bookmarks into tweets, threads, or long-form content

Best practices

• Install bird-cli and verify with bird whoami for the fastest workflow; otherwise set up X API v2 OAuth
• Start with -n 20 for quick digests; use --all only for deep analysis
• Store state (lastSeenId) so scheduled digests only surface new bookmarks
• Enable thread context (--include-parent) for replies to understand conversations
• Treat bookmarks older than a configurable threshold (default 30 days) as candidates for action or cleanup

Example use cases

• Daily cron digest that fetches new bookmarks, categorizes them, and emails a short action-first summary
• Content recycling: find high-like tweets in bookmarks and generate drafts in your voice for posting
• Research assist: group bookmarks about a trend (AI, crypto) and propose a short research plan or article outline
• Cleanup sweep: list stale bookmarks with TL;DR and offer to unbookmark them via bird CLI
• Ad-hoc analysis: search bookmarks for a keyword and return clustered results with recommended next steps