---

name: bookmark-processor description: This skill should be used when the user asks to "process bookmarks", "run /process-bookmarks", "analyze my twitter bookmarks", "find skill candidates from bookmarks", or wants to fetch, categorize, and extract insights from Twitter bookmarks for skill discovery. version: 1.0.0

Twitter Bookmark Processor

Automate the bookmark-to-skill discovery pipeline. Fetch, analyze, extract, propose, track, clean.

The Workflow

Fetch Bookmarks → Categorize → Score → Extract Resources → Generate Proposals → Update CSV → Unbookmark

How to Use

Process All Recent Bookmarks

/process-bookmarks

Process with Options

/process-bookmarks --keep-bookmarks    # Don't unbookmark after processing
/process-bookmarks --min-relevance 4   # Only process relevance 4+
/process-bookmarks --category AI-LLM   # Only process specific category

What Gets Processed

Categories

Category	Examples
AI-LLM	Claude tips, agent patterns, MCP servers
DevOps	CI/CD, infrastructure, deployment
Marketing	Growth tactics, content strategy
Business	Pricing, sales, startup advice
Design	UI/UX, visual design
Personal	Health, productivity, life advice
Other	Everything else

Relevance Scoring (1-5)

Score	Meaning
5	Definitely a skill - actionable workflow or pattern
4	Likely a skill - useful technique or tool
3	Maybe a skill - interesting but needs exploration
2	Not a skill - just information
1	Not relevant - personal/off-topic

Skill Candidate Criteria

A bookmark becomes a skill candidate when:

• Relevance score is 4-5
• Contains actionable workflow
• Has reusable pattern
• Includes tools/techniques
• Could benefit multiple projects

Output Files

CSV Tracking File

Location: ~/projects/adam/twitter-bookmarks.csv

Columns:

• handle - Twitter username
• tweet_text - Tweet content (truncated)
• url - Tweet URL
• timestamp - When bookmarked
• category - Assigned category
• relevance - Score 1-5
• plugin - Yes/Candidate/blank

Skill Proposals Document

Location: ~/projects/adam/skill-proposals-YYYY-MM.md

Contains:

• Summary table of candidates
• Detailed skill specifications
• Implementation notes
• Source attribution

Processing Steps

1. Fetch Bookmarks

Uses Twitter API to get recent bookmarks with:

• Tweet text (including long tweets)
• Author info
• URLs/links
• Engagement metrics

2. Categorize

Analyzes tweet content to assign:

• Category (AI-LLM, DevOps, etc.)
• Relevance score (1-5)
• Skill candidate flag

3. Extract Resources

For skill candidates, fetches linked resources:

• Blog posts
• GitHub repos
• Documentation
• Videos (extracts descriptions)

4. Generate Proposals

For each skill candidate:

• Summarizes core concept
• Identifies key features
• Notes implementation approach
• Estimates complexity

5. Update CSV

• Adds new entries
• Updates existing entries if re-bookmarked
• Maintains chronological order

6. Unbookmark Processed

Removes processed bookmarks from Twitter to:

• Keep bookmark list fresh
• Prevent reprocessing
• Signal "already captured"

Configuration

Set in CLAUDE.md or Memory

## Bookmark Processing Preferences

CSV Location: ~/projects/adam/twitter-bookmarks.csv
Proposals Location: ~/projects/adam/skill-proposals-{date}.md
Auto-unbookmark: true
Min Relevance for Skill: 4
Focus Categories: AI-LLM, DevOps

Example Session

User: /process-bookmarks

Claude:

Fetching bookmarks...
Found 25 new bookmarks since last processing.

Categorizing...
- AI-LLM: 15
- DevOps: 5
- Marketing: 3
- Personal: 2

Scoring relevance...
- Score 5: 8 bookmarks
- Score 4: 6 bookmarks
- Score 3: 7 bookmarks
- Score 2: 3 bookmarks
- Score 1: 1 bookmark

Skill candidates identified: 6

Extracting linked resources...
- Fetched 4 blog posts
- Fetched 2 GitHub repos

Generating skill proposals...
Created: skill-proposals-2026-01.md

Updating CSV...
Added 25 entries (total: 154)

Unbookmarking processed items...
Removed 25 bookmarks from Twitter.

Summary:
- New entries: 25
- Skill candidates: 6
- Resources extracted: 6
- Proposals generated: skill-proposals-2026-01.md

Integration with Other Skills

With skill-creator

When ready to implement a proposed skill:

/create-skill from proposal "Plain English E2E Testing"

With voice-learning

Extract voice patterns from thought leader bookmarks:

/learn-voice from bookmarks by @bcherny

Weekly Automation

Consider scheduling weekly processing:

# In cron or automation
claude -p "/process-bookmarks --quiet" >> ~/logs/bookmark-processing.log

Handling Rate Limits

If Twitter API rate limits hit:

• Processing pauses automatically
• Resume with /process-bookmarks --continue
• Partial progress saved to CSV

Privacy Note

• Bookmarks are private Twitter data
• Processing happens locally
• CSV is local file
• Unbookmarking is optional

Credits

Meta-skill developed during Twitter bookmark analysis session on 2026-01-08.