stenolisted

# steno — human-facing terse text Audience: reviewer scanning prose, wants facts fast — still human, ≠ token-optimised model. Readable symbols ∧ plain words; math glyphs → `glyph` skill ∈ `sdd` plugin. ## SKIM TEST Top criterion: a reviewer scans the text ∧ the fact appears ∈ the first sentence ∨ bold-lead bullet of each block. Skim fails → rewrite the first clause; ⊥ retain dense form, ⊥ cut further words. Self-check ∀ paragraph ∨ bullet: - First clause states the fact (subject + verb ≤ 8 words). - Subject ∧ verb both visible — symbol-chain fragments fail the test. - Anti-test: cover everything after the first clause; the core fact remains readable. Compression subordinate to this test — a word that aids the skim stays. ## SCOPE Criterion: human-facing terse prose ∀ non-author reviewers — scan facts, benefit from compression w/o math-glyph load. Common applications (⊥ exhaustive): - GitHub issues ∧ PRs — titles, bodies (incl. PR desc refresh on merge). - PR squash/merge commit message bodies (release-note section). - Insights comments emitted by gh skills. - READMEs ∧ user-facing docs where compression aids scan. ⊥ apply to: - Code, snippets, backticked text. - Conventional Commits title prefix (`type(area):`) — fixed format. - Error strings, log lines. - External-facing copy (marketing, landing pages). ## SENTENCE SHAPE Four rules ∀ prose sentence ∨ bullet body: 1. **Lead-first** — subject + verb open the sentence; topic-shift ∧ qualifier clauses move to the