Simplified Technical Hebrew (based on ASD-STE100)

ASD-STE100 is a controlled-language standard for English. The aerospace and defense industry built it so maintenance technicians cannot misread instructions. It removes the two biggest sources of misreading. These are words with more than one meaning, and sentences with more than one possible structure.

This skill applies the same discipline to Hebrew. Hebrew adds its own sources of misreading. Words can be unvocalized homographs, and prefix letters are glued to words. Construct chains (סמיכות) and chains of ש- clauses grow long. An instruction can take four competing forms. One reader is an AI agent that parses Hebrew with no human to ask. The other is a human reader of technical Hebrew, including readers with weaker Hebrew.

ASD-STE100 defines English only. Every Hebrew rule below is this skill's adaptation, not an official standard.

When to Use This Skill

• Hebrew text reads as dense, bureaucratic, or ambiguous. Examples: a tool description, an error message, an instruction between agents, a system prompt, a procedure, UI text.
• Hebrew text goes to an agent, a translation pipeline, or a reader with weaker Hebrew, and misreading has a cost.
• The user asks for a before/after table that names each broken rule. The default output is the rewritten text alone (see Output Format).

Do not use it for creative, marketing, or persuasive copy. Do not use it for English text: the English skill asd-ste100 covers English.

Two Modes

Pick a mode before you rewrite. If the user does not say which, infer it from the type of text. Keep the choice internal unless the user asks for the rule table.