MMind

This lightweight research and writing skill is based on information sources from the past 3 days. It extracts anchor points from user information sources and recommends 5 candidate topics (3 consensus hotspots + 2 forward-looking signals). After the user selects a topic, it delves into the relevant information and generates an editable outline. The main writing only begins after the user explicitly confirms the outline.

After inputting an industry, track, sub-sector, product type, or research topic, the system systematically discovers high-value, continuously updated, cross-platform, and cross-country/regional core information sources in that field, and compiles them into a list of information sources (human-readable table + agent-readable JSON) suitable for subsequent access, retrieval, and monitoring by the agent.

It helps creators transform existing aesthetic references into reusable, page-based narrative JSON style specifications through multiple rounds of verification.

This software converts reference content such as articles, reports, notes, Markdown, and extracted text from PDFs into A4 print-quality single-file HTML files. It adaptively selects accent colors based on the content theme, embeds semantic minimalist hand-drawn line art SVG, automatically fixes printing problems such as blank pages, orphaned lines, and truncated elements when exporting PDFs, and opens the PDF directly with a canvas panel in the final step, allowing users to view, print, or export PDFs without copying and pasting code.