skill-progressive-disclosure-pattern-for-oversized-files
When a skill SKILL.md exceeds the 600-line hard limit (setup-curator threshold), extract subsections to a docs/ subdirectory within the skill folder and replace extracted content with pointer blocks. This preserves every unique content line while reducing the main file under the 400-line target. The frontend-intelligence skill was split 761→406 lines across 6 docs files (792 total lines), with all unique content preserved.
Related
- skills-over-600-lines-split-via-progressive-disclosure
- docker
- 2026-04-04-oracle-001-self-architecture-analysis
- snowflake-v2-progressive-disclosure-for-large-skill-files
- snowflake-v2-skill-pattern-reduces-entry-point-token-cost-95
- skill-file-progressive-disclosure-for-over-600-line-splits
- progressive-disclosure-resolves-ambiguity-without-content-lo
- progressive-disclosure-resolves-structural-ambiguity-in-skil
- helix-skill-400-line-hard-limit-enforcement
- progressive-disclosure-split-grows-total-skill-footprint