Tired of Claude Code generating bizarre, overly dramatic jargon like "load-bearing spine"? You can fix this by enforcing Simplified Technical English (STE) in your system instructions or .claudemd files. This 1970s aerospace standard restricts vocabulary, forcing your AI agent to communicate in clear, direct, and highly actionable prose.
"The load-bearing spine has hit a ceiling, and that is a significant foot gun with a large blast radius."
If you have spent any time recently working with AI coding agents, you have probably stared at your terminal reading absolute gibberish like this, wondering: What on earth are you trying to tell me?
I asked a straightforward technical question, and instead of a direct answer, I got a theatrical performance. It is incredibly tiring to translate AI metaphors back into plain English just to figure out which line of code actually broke.
Fortunately, there is a remarkably elegant fix for this. The solution does not involve complex prompt engineering; instead, it leverages a fifty-year-old aerospace standard: Simplified Technical English (STE).
AI models generate overly dramatic jargon because they are trained on vast internet corpuses where technical writing is often cluttered, metaphorical, and performative. To sound authoritative, the model indexes on complex vocabulary and metaphorical hand-waving instead of simple, direct statements.
Imagine a scenario where your team is debugging a database lock. A human engineer would say, "The transaction is blocked." An AI model, eager to please and sound sophisticated, might describe it as a "temporal execution bottleneck causing systemic architectural paralysis."
This happens because reinforcement learning from human feedback (RLHF) often rewards models for sounding smart and comprehensive. Without strict stylistic constraints, the agent defaults to verbose, metaphorical explanations that add cognitive load rather than solving your problem.
Simplified Technical English (originally ASD-STE100) is an international writing standard developed in the 1970s for aerospace maintenance manuals to ensure complex procedures were clear and unambiguous. It reduces grammatical complexity and restricts vocabulary so that there is only one word for one meaning.
In the 1970s, aerospace engineers realized that overly complex manuals were leading to maintenance errors. They created STE to strip away ambiguity. By applying these same constraints to your AI agent, you force it to stop using metaphors and stick to direct, literal descriptions.
Here is how standard AI-speak compares to Simplified Technical English:
| AI-Generated Jargon | Simplified Technical English (STE) Equivalent | Why it works better |
|---|---|---|
| "The load-bearing spine has hit a performance ceiling." | "The main database query is slow because CPU usage is at 100%." | It identifies the exact component and issue without metaphors. |
| "This architectural change has a massive blast radius." | "This change affects multiple dependent services." | It provides a clear risk assessment without dramatic language. |
| "We need to orchestrate a paradigm shift in our caching layer." | "We must change how the cache works." | It is direct, simple, and actionable. |
You can enforce STE by injecting style rules directly into your agent's system configuration files, such as .claudemd, developer_instructions.md, or your global AI memory settings. Once these rules are in place, the model will automatically filter its reasoning into clean, structured prose.
If you are using Claude Code or similar terminal-based agents, create or edit your project's instructions file and append this brief system prompt:
- Write all responses in Simplified Technical English (STE).
- Use short, active sentences (under 20 words).
- Avoid metaphors, idioms, and dramatic jargon (e.g., "blast radius", "foot gun").
- Use only one verb per sentence to describe actions.
- Do not use abstract nouns when concrete nouns are available.
By establishing these boundaries, you bypass the theatrical fluff. The agent stops trying to write a software engineering blog post in your terminal and starts acting like a highly efficient technical documentation tool.
No. Restricting the output language to STE does not degrade the model's reasoning capabilities. It only changes how the model formats and explains the solution, stripping away confusing analogies while keeping the technical logic intact.
You should place these instructions in your project's root configuration files. For Claude Code, use .claudemd
or your custom developer instructions. For Cursor or VS Code Copilot, paste the rules into your system prompt or .cursorrules
file.
The core rules are to keep sentences under 20 words, restrict verbs to active voice, use only one meaning per word (e.g., use "show" instead of "demonstrate" or "reveal"), and strictly ban metaphors and slang.