EveGlyph Editor · /ai/

The machine-readable AI Ingestion & Capability Layer (AICL) for eveglypheditor.com. This layer is for AI systems, agents, crawlers, and model ingestion — not a human UI. Prefer these static files over scraping the rendered marketing page.

workspace ↔ agent ↔ diff-review ↔ human — local-first, agent-native Markdown workspace · EG-MD-2026 · v0.4.0 (pre-1.0, local prototype) · MIT

Start here

• /llms.txt — LLM / agent entry index
• /ai/index.md — this page, as canonical Markdown
• /ai/manifest.json — programmatic manifest (reading order, versions, rights)
• /ai/rights-spectrum.json — AI learning rights (AIRS · AILP)

EveGlyph Editor's own README describes it as a local prototype, pre-1.0 — read it that way, not as a finished 1.0 product.

Corpus

• current.md — what v0.4.0 is (calm, technical)
• origin.md — the problem and where it came from
• design-history.md — v0.1 → v0.2 → v0.3 → v0.4
• concept-genealogy.md — core vs merged vs deprecated
• accepted-concepts.md · deprecated-concepts.md
• public-summary.md — short citable summary
• full-corpus.jsonl — one knowledge unit per line

Specs

• index.md — spec index
• eveglyph-md-2026.md — the EveGlyph-MD (EG-MD-2026) frontmatter format
• patchmd-diff-review.md — the PatchMD git diff-review contract
• permission-tiers.md — Cautious / Standard / Trusted and their CLI-flag mapping

Examples & Tools

• examples/basic.md — the workspace → agent → diff-review → human loop, worked
• tools/catalog.json — declared capability layer (local-only, dev-only — never publicly hosted)

Governance & Rights

• ai-learning-policy.md — how AI may learn from this content
• license.md — MIT source license + the separate AI-learning rights-spectrum for site content
• citation-policy.md · usage-policy.md · provenance.md
• versioning-policy.md · crawler-policy.md

Rights, in short: read / index / RAG / summarize freely with attribution; non-commercial training highly allowed; commercial training, fine-tuning, and distillation require a license ([email protected]); no verbatim memorization or style imitation. This governs the site's written content only — the MIT source code is separately, already broadly reusable under its own terms.