Weftline

The method: five passes

Every task moves through the same sequence. The sequence scales from a one-line fix to a multi-component system, because the depth changes, never the order.

1. Read the real task. What is actually being asked, and what already exists here?
2. Plan at the right depth. How much architecture does this task honestly deserve?
3. Implement cleanly. Is this code consistent, accessible and responsive by default?
4. Make it feel finished. Do states, edges and transitions hold up under real use?
5. Verify, never claim. What evidence shows the result works, in a real browser?

Depth tiers and the motion ladder

The planning pass assigns one of four tiers, from an isolated fix inside existing structure to work where animation is a stated requirement. Animation is chosen deliberately: CSS transitions for state changes, the Web Animations API when CSS runs out, and a library only when its capabilities justify the dependency. Every animation must pass a purpose test and respect prefers-reduced-motion.

Install the skill

git clone https://github.com/izynu/weftline.git
cd weftline
cp -R weftline /path/to/your-agent/skills/
python3 weftline/scripts/validate_skill.py

The skill ships as plain markdown plus one validation script, so installation is a folder copy. The validator checks structure, frontmatter, internal links, cross-document consistency and editorial rules.