AI infographic and layout tools (SenseNova & infocard-skills)

Infocard-skills and SenseNova automate layouts and infographic editing. - infocard-skills: open-source Agent Skill - Generates magazine-grade HTML info cards - Auto judges information density and picks layout skeleton - Renders to 3:4, 16:9 and other aspect ratios - Six-step workflow with strict quality redlines - Quality rules: title preserved, no large blanks, module hierarchy - SenseNova U1 was open-sourced by SenseTime - SenseNova V3 adds infographic editing capabilities - High-precision local text and global style edits - Edited text auto-adapts to original design and typography - Model weights, docs and community links provided - Future: autonomous iterative post-editing and self-correction
Key insights
infocard-skills produces magazine-grade info cards
infocard-skills is an open-source Agent Skill that ingests natural-language content, judges information density, selects a layout skeleton in a modern magazine/Swiss Internationalism style, and generates finalized HTML info cards that render to multiple aspect ratios (3:4, 16:9, etc.). The skill implements a six-step workflow with strict quality redlines (title must not be rewritten, no large blank areas, modules must have hierarchy) and runs automatic post-render checks to enforce layout and quality rules.
SenseNova V3 adds high-precision infographic editing
SenseNova, an open-source project by SenseTime, released an enhanced V3 that introduces infographic editing capabilities, including high-precision local text and overall style adjustments. The system can adapt edited text to the original image’s design and typography, allowing users to preserve core information while changing visual expression across brands or themes. V3 includes model weights, documentation and community links for download and use.
Autonomous iterative post-editing planned
Both projects point toward autonomous iterative refinement: infocard-skills performs render-time self-checks; SenseNova V3 outlines next-step capabilities where the model will check generated infographics, call its own editing abilities to fix issues, and run multiple iterations to continuously optimize the final image.