Bookmark Income Engine

A Claude Code plugin (Cowork-compatible) that transforms your X (Twitter) bookmarks into scored revenue opportunities, skill upgrades, and 30-day implementation plans using a multi-agent pipeline.

What It Does

Run /process-bookmarks and the engine will:

1. Capture bookmarks from X via browser automation (no API key needed)
2. Classify each bookmark as Opportunity, Skill, Hybrid, or Low Priority
3. Research opportunities with web search, validate revenue claims, and score plausibility
4. Map skills from bookmarks to your existing projects with concrete action items
5. Generate 30-day plans for GO-rated opportunities with cost breakdowns and revenue projections
6. Output everything to a Notion database and local markdown

Architecture

/process-bookmarks [count]
        |
   bookmark-capture (skill)        <- browser automation, dedup
        |
   bookmark-orchestrator (agent)   <- classify, dispatch, merge, rank
        |
   +--------+--------+
   |                  |
opportunity        skill-compiler  <- parallel specialist agents
-analyst           (agent)
(agent)      
   |
implementation-planner (agent)     <- 30-day plans for GO opportunities
        |
   Notion DB + markdown output

Agents

Agent	Model	Role
bookmark-orchestrator	Opus	Pipeline coordinator -- classifies, dispatches, merges, ranks, writes output
opportunity-analyst	Opus	Deep research, revenue validation, plausibility scoring, GO/NO-GO recommendation
implementation-planner	Opus	30-day execution roadmaps for opportunities with plausibility >= 7
skill-compiler	Sonnet	Extracts techniques/tools and maps them to existing projects

Skills

Skill	Purpose
process-bookmarks	Main entry point -- orchestrates the full pipeline
bookmark-capture	Browser automation to read and parse X bookmarks

Requirements

• Claude Code desktop app or CLI
• Claude in Chrome extension -- installed and connected to your Claude Code session
• Logged into X (Twitter) in Chrome
• Notion MCP server connected (optional -- for database output; markdown output works without it)

Installation

1. Create a Cowork Project

Open Claude Code (desktop app or CLI) and create a new project, or open an existing one where you want bookmark processing available.

2. Add the Plugin

In your Cowork project, go to Customizations (the + next to "Personal plugins") and choose one of:

• Add marketplace -- enter the GitHub repo WillyV347/twitterBookmarks to register the marketplace, then enable the Bookmark Income Engine plugin
• Upload plugin -- clone this repo locally and point to the plugins/bookmark-income-engine directory

Once added, the /process-bookmarks skill and all 4 agents will appear under your project's Skills and Agents.

3. Configure Your Profile

The plugin uses a config.yaml file to personalize scoring and project mapping. Without it, the engine won't know what opportunities are relevant to you.

Save config.yaml OUTSIDE the plugin install directory. The plugin bundle is often mounted read-only (Cowork / plugin-mount contexts), so a config saved inside it will not persist between sessions.

1. Choose a stable location outside the plugin. Recommended: ~/Documents/Claude/Projects/AI Projects/bookmark-income-engine/config.yaml (or any path you set via the $BOOKMARK_ENGINE_CONFIG environment variable)
2. Copy the template there: cp config.example.yaml "$HOME/Documents/Claude/Projects/AI Projects/bookmark-income-engine/config.yaml"
3. Edit that config.yaml with your:
   • Mission statement -- bookmarks are scored against this lens
   • Project portfolio -- your existing projects (name, path, tech stack, revenue status)
   • Skills profile -- your languages, platforms, and strengths
   • Output directory -- where markdown archives are saved

The skill resolves the config in this order: $BOOKMARK_ENGINE_CONFIG, then the recommended stable path above, then <plugin-root>/config.yaml as a last-resort fallback (which will not persist). The config file is gitignored so your personal data stays local.

4. Connect Required Tools

Make sure these MCP servers are connected in your project:

• Claude in Chrome -- for browser automation (bookmark capture)
• Notion -- for database output (optional but recommended)

5. Set Up a Scheduled Task

The real power is running this on autopilot. In your Cowork project, create a scheduled task:

1. Go to Scheduled Tasks (or use /schedule)
2. Create a new task with a name like bookmark-triage
3. Set a cron schedule (e.g., twice a week: 0 9 * * 1,4)
4. Set the prompt:

   Run /process-bookmarks 30. After completion, summarize the top 3 opportunities.
   

The engine will run on schedule, capture your latest bookmarks, research opportunities, and populate your Notion pipeline -- no manual intervention needed.

Usage

Once installed and configured, run manually anytime:

/process-bookmarks        # Process up to 30 bookmarks (default from config)
/process-bookmarks 50     # Process up to 50 bookmarks

Or let your scheduled task handle it automatically.

Output

• Notion: "Bookmark Income Pipeline" database with board, table, and list views
• Markdown: output/run-YYYY-MM-DD.md -- full run archive
• Dedup tracker: output/.last-run -- prevents reprocessing bookmarks across runs

Project Structure

plugins/bookmark-income-engine/
  .claude-plugin/plugin.json     # Plugin manifest
  config.example.yaml            # Configuration template (tracked)
  config.yaml                    # Your personal config (gitignored)
  agents/
    bookmark-orchestrator.md     # Pipeline coordinator
    opportunity-analyst.md       # Revenue opportunity researcher
    implementation-planner.md    # 30-day plan generator
    skill-compiler.md            # Skill-to-project mapper
  skills/
    process-bookmarks/SKILL.md   # Main entry point skill
    bookmark-capture/SKILL.md    # Browser automation capture
  commands/
    process-bookmarks.md         # Command alias for the skill
  scripts/
    last-run-tracker.sh          # Dedup tracker utility
  output/                        # Run archives (local)

License

MIT