Problem
phonosemantics.md is 183 lines and loaded in its entirety during Step 3. But most naming sessions only need the summary tables (consonant types, vowel positioning) — not the full phonosemantic clusters section or the practical application checklist.
Fix
Split into two files:
- phonosemantics.md (core, ~80 lines) — keep the consonant types, vowel positioning, and rhythm/stress tables. This is what Step 3 needs.
- phonosemantics-detail.md (~100 lines) — move the phonosemantic clusters table, matching guide, and practical application checklist here. Load this only during Step 6 (Evaluate) when doing detailed sound analysis.
Update SKILL.md to load the core file in Step 3 and the detail file in Step 6.
Why this matters
Every line loaded into context costs tokens. Most naming sessions don't need to know that "gl-" connotes light or that "sn-" connotes nose movement. They need the high-level consonant/vowel guidance. Save the detail for evaluation.
Before submitting
Read CONTRIBUTING.md and run `markdownlint-cli2 '**/*.md'` before pushing.
Scope
Split one file into two + update references. Medium PR.
Problem
phonosemantics.md is 183 lines and loaded in its entirety during Step 3. But most naming sessions only need the summary tables (consonant types, vowel positioning) — not the full phonosemantic clusters section or the practical application checklist.
Fix
Split into two files:
Update SKILL.md to load the core file in Step 3 and the detail file in Step 6.
Why this matters
Every line loaded into context costs tokens. Most naming sessions don't need to know that "gl-" connotes light or that "sn-" connotes nose movement. They need the high-level consonant/vowel guidance. Save the detail for evaluation.
Before submitting
Read CONTRIBUTING.md and run `markdownlint-cli2 '**/*.md'` before pushing.
Scope
Split one file into two + update references. Medium PR.