glyphlisted

# glyph — LLM-facing math-glyph encoding Audience: the LLM reading SPEC.md back into context. SPEC.md = LLM-facing artifact — ∀ reads/writes via Claude. humans operate through /sdd:* cmds, ⊥ hand-edit; /sdd:explain decodes glyph → prose ∀ human consumption. ∴ math glyphs ∧ fragments = features ⊥ bugs — model parses them cheaply, compression aggression yields net token savings on every re-load. "Math-glyph" ≠ "glyph". Generic glyphs are any printable symbols (status markers `.` `x`, typographic dots, bullets). Math-glyphs are the specific math operators below (∀ ∃ ∴ ≡ ⊥ ¬ ∈ ∉ ≤ ≥ ≠ → ∧ ∨ §) — that's what makes this encoding compress so hard, and why a separate `steno` skill exists for human-facing text where math-glyphs would slow reviewers down. Applies to SPEC.md writes, spec-referencing prose, backprop entries. Does NOT apply to code, error strings, commit messages, PR descriptions, or anything a human reviewer reads on GitHub (use the `core:steno` skill for that). ## GRAMMAR - Drop articles (a, an, the). - Drop filler (just, really, basically, simply, actually). - Drop aux verbs where fragment works (is, are, was, were, being). - Drop pleasantries. - No hedging (skip "might", "perhaps", "could be worth"). - Fragments fine. - Short synonyms per `## VERBS` (canonical verbs + `avoid` column). - Pipe tables compact: ⊥ alignment padding, ⊥ separator row. Header row + data rows, bare `|cell|cell|` (escape literal `|` as `\|`). ## SYMBOLS Prefer over words: ``` → leads