SimpleEnglish: Using Aerospace Standards to Fix AI Writing

SimpleEnglish is a tool that forces AI agents to follow the ASD-STE100 Simplified Technical English standard used in aerospace manuals. It eliminates wordy 'AI slop' by enforcing 53 strict rules on sentence structure, voice, and vocabulary. Testing proves it reduces writing errors by over 70% while making AI-generated documentation shorter and easier to understand.

Key Points

• LLMs often produce 'AI slop' that is wordy and ambiguous, which SimpleEnglish fixes by enforcing the ASD-STE100 aerospace writing standard.
• The skill implements 53 specific rules, including strict sentence length limits, active voice requirements, and the removal of hedging language like 'should' or 'might'.
• Empirical testing shows a 72.9% reduction in writing violations and shorter overall output across multiple AI models.
• The tool is designed for technical use cases such as documentation, error messages, runbooks, and incident reports where clarity is critical for safety.
• It is highly accessible, supporting the Agent Skills standard for CLI tools and providing system prompts for web-based chatbots like ChatGPT and Gemini.

Sentiment

The community is mildly positive toward the underlying idea but skeptical of the specific packaging. Commenters generally agree that ASD-STE100 and related style constraints can improve AI-generated technical writing, yet many doubt that this needs a full agent skill or that it fully solves AI slop. The tone is practical and critical rather than dismissive, with the strongest disagreement aimed at overclaiming, context overhead, and unsafe rewriting examples.

In Agreement

• Controlled technical English can make LLM output shorter, clearer, less corporate, and easier to edit.
• Adding ASD-STE100 guidance to prompts, assistant profiles, or project instruction files can noticeably reduce verbose AI-writing habits.
• The post is valuable because it introduces many readers to a mature controlled-language standard they can reuse in their own workflows.
• Style constraints for agent-facing documentation are useful because agent instructions behave more like interfaces than ordinary prose.
• STE's focus on clear instructions for non-native speakers and safety-critical contexts is relevant to technical documentation generated by AI.

Opposed

• A dedicated agent skill may be unnecessary because a short system prompt can ask for ASD-STE100 or concise technical English directly.
• The project risks wasting context window space with rules that models may already know from training data.
• Simplified Technical English is not a magic fix; good controlled-language writing still requires domain knowledge and careful editing.
• The repository's README and examples contain the same AI-writing tells the project claims to remove, which weakens trust in the tool.
• Some examples appear to add unsupported operational details or promises, making the tool potentially unsafe if used as a blind rewrite step.
• External linters, pre-commit hooks, or iterative validation may be more reliable than asking an LLM to obey a style standard unaided.