Vectorscope 2.112 — fewer tokens, fewer errors

• Agent instructions rewritten in controlled technical English (ASD-STE100): rules, procedure, verification checklist.
• SKILL.md: −73% tokens (≈10,900 → ≈2,900). Details moved to references/, read only when needed.
• Shorter description in the agent's context: 944 → 632 characters.
• Measured on a full agent run, page to report: 114 → 82 s, script errors 1 → 0.
• Fix: the lost-FAQ finding shows its counts.
• Fix: a passage that opens on a lower-case brand (pdfFiller, iPhone) is no longer flagged as mid-sentence.

Vectorscope 2.111 — how generated the text sounds, and the report in Polish

• Shows how an AI retriever reads a page: fetch, text, passages, the 26-passage cap.
• robots.txt verdicts for AI crawlers and automatic structure checks.
• The agent groups passages, tests queries and writes a rewrite plan.
• How generated the text sounds: Pattern score, marker words, swap test.
• Report in English, Ukrainian and Polish, light and dark.
• The scripts call no model and no embeddings.