cd /news/ai-agents/agents-misread-hebrew-a-40-year-old-โ€ฆ ยท home โ€บ topics โ€บ ai-agents โ€บ article
[ARTICLE ยท art-155137] src=dev.to โ†— pub= topic=ai-agents verified=true sentiment=โ†‘ positive

Agents misread Hebrew. A 40-year-old aerospace standard helps.

A developer ported the ASD-STE100 Simplified Technical English skill to Hebrew, releasing g4lb/asd-ste100-hebrew-skill as an MIT-licensed Claude Code skill for making Hebrew tool descriptions and error messages easier for AI agents to parse. The port adapts STE100 rules that define only English, adding Hebrew-specific guidance such as keeping ืืช/ืžืฉืชืžืฉ as a noun and splitting hedged error messages into separate claims so uncertainty is preserved. A standard-library Python linter, scripts/ste-lint-he.py, performs a mechanical first pass on semicolons, sentence length, nominalization and marketing words, while construct chains, homographs and dropped ืืช remain for the model and author to check.

by read2 min views1 publishedOct 22, 2026

In aerospace maintenance, a misread instruction can ground an aircraft, or cause worse damage. So the industry built ASD-STE100, Simplified Technical English. It is a controlled language with one meaning per word, active voice, short sentences, and one instruction per sentence.

Today a new kind of reader parses technical text with no one to ask: the AI agent. A model reads tool descriptions, error messages, and instructions between agents. It cannot ask the author what the author meant. danyuchn/asd-ste100-skill applies STE100 to English agent text as a Claude Code skill. I ported it to Hebrew.

STE100 defines English only. No official Hebrew version exists, so every Hebrew rule here is an adaptation. Some English rules carry over directly. Others have no English equivalent:

ืืช.ืžืฉืชืžืฉ is both "user" and "uses". Keep it a noun and use a different verb: "ื”ืžืฉืชืžืฉ ืžืคืขื™ืœ ืืช ื”ื›ืœื™".ืœื‘ืฆืข with an action noun. Before:

ื™ื™ืชื›ืŸ ืฉืื™ืจืขื” ืฉื’ื™ืื” ื‘ืžื”ืœืš ืขื™ื‘ื•ื“ ื”ื‘ืงืฉื” ืฉืœืš ืขืงื‘ ืื™-ื”ืชืืžื” ืืคืฉืจื™ืช ื‘ืคื•ืจืžื˜ ื”ื ืชื•ื ื™ื ื”ืฆืคื•ื™, ืืฉืจ ืขืฉื•ื™ื” ืœื”ื™ื’ืจื ื›ืชื•ืฆืื” ืžื’ืจืกืช ืœืงื•ื— ืฉืื™ื ื” ืžืขื•ื“ื›ื ืช.

After:

ื™ื™ืชื›ืŸ ืฉืื™ืจืขื” ืฉื’ื™ืื” ื‘ืขื™ื‘ื•ื“ ื”ื‘ืงืฉื”. ื™ื™ืชื›ืŸ ืฉื”ืกื™ื‘ื” ื”ื™ื ืื™-ื”ืชืืžื” ื‘ืคื•ืจืžื˜ ื”ื ืชื•ื ื™ื ื”ืฆืคื•ื™. ื™ื™ืชื›ืŸ ืฉื’ืจืกืช ืœืงื•ื— ืœื ืžืขื•ื“ื›ื ืช ื’ืจืžื” ืœืื™-ื”ื”ืชืืžื” ื”ื–ื•.

"ืื™ืจืขื” ืฉื’ื™ืื”" ("an error occurred") reads better. But it states an error that the system only suspects. A simplifier that removes uncertainty makes the text clearer and wrong. The skill splits the sentence into three claims and keeps every hedge.

A Python linter (scripts/ste-lint-he.py) uses only the standard library. It does a mechanical first pass: semicolons, sentence length, nominalization, marketing words, prefix hyphens, slang, and synonym rotation. It cannot check construct chains, homographs, or a dropped ืืช. The model and the author check those.

npx skills add g4lb/asd-ste100-hebrew-skill

Then ask Claude: ืคืฉื˜ ืืช ืชื™ืื•ืจ ื”ื›ืœื™ ื”ื–ื” ("simplify this tool description").

Repository (MIT license): https://github.com/g4lb/asd-ste100-hebrew-skill

If you write Hebrew for agents, tell me which rules feel wrong.

โ”€โ”€ more in #ai-agents 4 stories ยท sorted by recency
โ”€โ”€ more on @asd-ste100 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain โ€” perfect for shipping the agent you just read about.

$git push zahid main
โ†’ Live at https://your-agent.zahid.host โœ“
Get free account โ†’ Pricing
from โ‚ฌ0/mo ยท no card required
LIVE [news/agents-misread-hebreโ€ฆ] indexed:0 read:2min 2026-10-22 ยท โ€”