reader-deep-dive

⚠Review·Scanned 2/18/2026

This skill generates a WhatsApp "Deep Dive" by reading Readwise saves, searching the user's archive, synthesizing a briefing, and sending it via clawdbot. It requires READWISE_TOKEN and TARGET_NUMBER, executes scripts/brief.sh which calls curl, gemini, and clawdbot, and queries https://readwise.io/api/v3/list/?location=new&page_size=10.

from clawhub.ai·v2aa7039·9.5 KB·0 installs
Scanned from 0.1.0 at 2aa7039 · Transparency log ↗
$ vett add clawhub.ai/sameerbajaj/reader-deep-diveReview findings below

Reader Deep Dive 🤿

Your reading list shouldn't be a write-only memory.

Most of us save articles, read them (maybe), and then forget them. We have a library of thousands of ideas, but we treat every new day like we're starting from zero.

This agent changes that. It scans what you're reading right now and connects it to the deepest cuts from your past. It doesn't just list links; it builds a timeline of your thinking on a specific topic.

What It Does

Every afternoon (or on-demand), it runs a comprehensive research pass:

  1. Detects Your Current Focus: It analyzes your last 24 hours of saves to identify the core theme—whether it's "AI Agents," "Vibe Coding," or "Metabolic Health."
  2. Mines Your Archive: It searches your entire history for that topic, pulling from thousands of documents saved over years.
  3. Synthesizes Intelligence: It selects at least 5 deep-dive connections and organizes them chronologically (Oldest to Newest) to show you the history of your thinking.
  4. Enriches Metadata: Every reference includes the Author Name and Media Type (tweet, video, article, etc.) for high-signal context.

The Output

You get a high-signal WhatsApp message formatted for mobile readability:

TOPIC: AI AGENTS 🎯

CURRENT FOCUS

  • Why OpenClaw feels alive (tweet) by claire vo (Saved: Today) Argues that "aliveness" comes from reactive loops rather than raw intelligence.

DEEP DIVE CONNECTIONS

  • One Year Ago: AutoGPT: The Dream of Autonomy (video) by Prompt Engineering (Saved: 2024-05-20) The original vision for self-improving agents. Revisiting this highlights the progress from "prompt engineering" to the "agentic engineering" patterns in your current saves. https://read.readwise.io/read/xyz

  • Last Month: The Rise of the Wrapper (article) by Nat Friedman (Saved: 2025-11-12) An analysis of LLM thin-clients. This provides the architectural context for the more complex stateful frameworks you are exploring today. https://read.readwise.io/read/abc

[... at least 3 more items ...]

Setup

  1. Get a Token: Grab your Readwise Access Token from readwise.io/access_token.
  2. Set Environment: 
    export READWISE_TOKEN="your_token_here"
  3. Run: 
    bash scripts/brief.sh

Requirements

  • curl, jq
  • clawdbot (for LLM inference and WhatsApp delivery)
  • gemini CLI (for stateless completion)