Sn Image Imitate

This is a three-stage pipeline that extracts a long caption from a reference image, rewrites it to match your new content request while locking down the layout, then generates the result. It's built on top of SenseNova's base image tools and enforces layout consistency through a post-generation review step with bounded retries. Use it when you want "make me a new slide deck page that looks like this one but talks about Q4 results instead of Q3," not when you need pixel-perfect style transfer or local edits. The two-tier agent architecture keeps the worker focused on the fixed pipeline while the main agent handles user interaction. Default is three attempts at 0.75 layout similarity threshold.

Install to Claude Code

npx -y skills add opensensenova/sensenova-skills --skill sn-image-imitate --agent claude-code