Blogtool

Overview

A multi-blog hub for running several blogs side by side. Each blog carries its own topic axis, platform, tone and length rules; the tool automates everything from topic discovery to draft writing, leaving paste and publish to the human.

It is a rebuild that generalizes an earlier tool (blog-kit) which was hard-wired to a single site — topic axes and tone rules lived in code, so adding a blog meant cloning the code. Blog profiles are now data (JSON) that can be registered and switched at runtime.

Highlights

• One-click pipeline — title selection → prompt build → draft → fact-check → image sourcing → machine validation, all behind a single button. No manual hand-offs between steps.
• Fact-check gate — figures and claims in the draft are checked against web sources, then corrected or removed, with the verdict reason and source quote kept in a report. Sentences with no traceable basis are cut.
• Trend-aware suggestions — web search is combined with the model to propose timely topics, and even the blog's own topic axis. Duplicates are flagged.
• Verify, then publish by hand — AI tells, length, link count and duplication are checked automatically, and the body exports through five paste modes. The final call stays human.

Stack

Python tkinter GUI (standard library only) + a Node.js CLI engine (no npm dependencies) + headless Claude CLI calls. Data is stored as JSON, split into dynamic/ (blogs, topics, ideas) and static/ (config, rules). Entirely local — no server, no database.