ASD-STE100 Skill

Overview

A Claude Code skill that rewrites dense, ambiguous English into ASD-STE100 Simplified Technical English. It is aimed at AI agents and non-native readers that must parse tool descriptions, error messages, inter-agent instructions, and other agent-facing English without a human to resolve ambiguity.

Key Features

• Applies ASD-STE100 principles to rewrite ambiguous agent-facing English.
• Offers a strict mode for procedures, error messages, and tool descriptions.
• Offers an STE-flavored mode for READMEs, PR descriptions, and explanatory prose.
• Flags rule violations such as ambiguous word choice, passive voice, complex tenses, long sentences, and dropped words.
• Includes a deterministic linter for structural patterns such as semicolons, soft phrasal verbs, nominalizations, marketing adjectives, passive voice, and present-perfect forms.
• Outputs rewritten text without preamble, mode announcement, or change summary.

Use Cases

• Rewriting agent-to-agent messages and inter-agent instructions.
• Clarifying tool and function descriptions for machine readers.
• Rewriting error messages and system prompts.
• Checking structural STE patterns in Markdown or text files with the linter.

Getting Started

• Run npx skills add danyuchn/asd-ste100-skill from your project root to install the skill for the current project.
• Clone the repository into ~/.claude/skills/asd-ste100 to make it available in every Claude Code project.
• Trigger the skill by asking Claude to disambiguate, rewrite, or apply ASD-STE100 to English text.

Before You Adopt

• License: MIT. Review its terms before using, modifying, or distributing the project.
• The skills CLI sends anonymous install telemetry.
• The linter checks only structural patterns and does not verify meaning preservation.
• The standard cannot be freely redistributed; reproduction requires ASD's written authority or an authorised organisation type.
• The repository's own prose does not pass the linter; findings are examples, not defects.
• The linter does not check noun-cluster length, phrasal verbs outside its list, hedges, or modality.