Overview

Every day I browse through countless technical docs and GitHub repos, but that exploration disappears the moment I close the tab. log-blog is a Python CLI tool that reads Chrome browsing history and automatically converts it into Hugo-compatible blog posts.

What Is log-blog?

ice-ice-bear/log-blog automates the “explore → organize → share” cycle. It extracts data from Chrome’s SQLite history database, fetches content from each URL using Playwright, converts it to Hugo-compatible markdown, and commits the result to the blog repository.

Pipeline Structure

graph TD
    A[Chrome History SQLite DB] -->|extract| B[URL + Title + Timestamp JSON]
    B -->|AI classify| C[Tech vs Non-tech]
    C -->|fetch| D[Enriched Content]
    D -->|AI write| E[Hugo Markdown Post]
    E -->|publish| F[Git Commit to GitHub Pages]

Step 1: Extract

log-blog extract --json --hours 24

Reads the recent N hours of visit history from Chrome’s SQLite history DB. Outputs URL, title, visit count, and last visit time as JSON.

Step 2: Classify

Integrated with the Claude Code skill system, AI classifies each URL as tech or non-tech, then groups them into YouTube, GitHub, and Docs/Web categories.

Step 3: Fetch

log-blog fetch --json "URL1" "URL2" "URL3"

Content is collected using a strategy appropriate to the URL type:

Step 4: Write & Publish

AI writes a technical blog post from the collected content, then the publish command commits it to the blog repository.

log-blog publish post.md        # Local commit only
log-blog publish post.md --push # Commit + push

Tech Stack

src/log_blog/
  cli.py             # CLI entry point (extract, fetch, publish)
  config.py          # YAML config loader
  history_reader.py  # Chrome SQLite history reader
  content_fetcher.py # Playwright-based content extractor
  post_generator.py  # Hugo markdown post generator
  publisher.py       # Git commit/push

• Python 3.12+ — primary language
• Playwright — browser automation for dynamic page content
• SQLite — direct access to Chrome’s history DB
• Claude Code Skill — AI classification, summarization, and writing

Configuration

chrome:
  profiles: ["Default"]
  history_db_base: "~/Library/Application Support/Google/Chrome"

time_range_hours: 24

blog:
  repo_path: "~/Documents/github/ice-ice-bear.github.io"
  content_dir: "content/posts"
  language: "auto"

playwright:
  headless: true
  timeout_ms: 15000
  max_concurrent: 5

Quick Links

• mindai/mega-code PR #26 — add_behavioral_validation (Upskill)
• mindai/megaupskill — MegaUpskill project

Insights

The core value of log-blog is “turning exploration itself into content.” The technical browsing that happens every day in a browser is already a learning process — but without capturing it, it vanishes. This tool automatically captures and structures that process. In fact, this very post was produced by that pipeline — Chrome history extraction → AI classification → content fetch → AI writing → blog deployment.