ASD-STE100 Agent Skill: Cutting LLM Slop in Terminal Outputs

X postpili 🪴Tool

A new open-source agent skill instructs language models to write in ASD-STE100 Simplified Technical English. Originally designed for aerospace documentation, the controlled language forces AI outputs to be concise, direct, and unambiguous.

Key points

  1. Developer @0xpili_ shared an agent skill configuring LLMs to write in ASD-STE100 to drastically reduce slop in terminal outputs.
  2. The GitHub repository provides 53 writing rules, a list of 869 approved words, replacement dictionaries, and a Python verification script.
  3. The skill installs into Claude via ~/.claude/skills/ or operates as part of the system prompt in other LLMs.
  4. The approach aligns with advice from Andrej Karpathy, who recommended ASD-STE100 prompts to enforce clean, constrained text generation.
  5. The Python tool identifies long sentences, passive voice, unapproved verbs, and unlisted words, though manual review remains necessary for official compliance.

Why it matters

Verbose model outputs clutter terminal workflows and slow down debugging. Enforcing a strict, controlled technical vocabulary removes fluff, passive voice, and conversational fillers, producing clear procedural statements that developers can parse instantly.

Analysis

ASD-STE100 is tailored for critical technical manuals where ambiguity causes errors. Implementing Issue 7 (2017) as an agent skill bridges formal documentation standards and modern LLM agents. While some community members suggested a simple prompt instruction might suffice, maintaining approved dictionaries and custom linter tools offers higher repeatability across repository workflows, even if token savings were not the primary goal.

Original post

pili 🪴

@0xpili_ · 3 October 2026

the amount of slop has been drastically reduced in my terminal since my agents use this skill to write in ASD-STE100: t.co/rfGI76n8L5 t.co/fngGvm2Zfv

3.3k likes59 replies6k bookmarks307.4k views

View on X
More posts in the thread (9)
  1. @interplato I haven’t measured token savings, my main objective was removing the slop
  2. @danigonlinea let me know if you try it!
  3. @PiyushDinde I think in the short term the skill is much better. In the long term, I am not sure
  4. @Mando_Ethan you can!
  5. @Mando_Ethan t.co/Eu9QsleAnc t.co/OqZeOEIVxX
  6. @illneil what do you mean?
  7. @illneil i don’t agree, you should try it and find it put yourself
  8. @stevenelliott let me know what do you think if you try it
  9. @FarzinNasirii I prefer to write myself rather than having an llm write for me so I don’t use any copywriting skills
Selected replies (5)
  • @Mando_Ethan @0xpili_ Why can't you just put a one liner in agents.md/claude.md to use ASD-STE100?
  • @DerMannCodes @0xpili_ I just added 'give responses in ASD-STE100' in my project root and it worked well. you didn't have to create this AI slop shit skill and increase token consumption.
  • @stevenelliott @0xpili_ This is life changing. No longer do I need to warp my mind to understand wtf it’s attempting to output.
  • @FarzinNasirii @0xpili_ Lovely. By the way do you know of any skill for copywriting? I find GPT models to be really really bad at this.
  • @PiyushDinde @0xpili_ Tried it. Very useful. Q for you, using this skill is objectively better than asking an agent to give you the o/p in ASD-STE?

Mentioned links

Links and tools from this entry

Summary generated by AI (Gemini 3.8 Flash, 5 October 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • Repository:0xpili/simplified-technical-english

    Simplified Technical English as an Agent Skill for LLMs

    An open-source project provides prompt guidelines and checking tools to help LLMs like Claude write and review technical documentation according to the ASD-STE100 standard.

    790starsPython

    AI & LLMs· Tool

  • Repository:troopdegen/mucho-texto

    mucho-texto: Structured Output Compression for AI Coding Agents

    mucho-texto is a skill designed for AI assistants like Claude Code, Cursor, and AGENTS.md workflows. It prevents agents from outputting verbose reports by categorising finished work into four distinct buckets and requiring an uncompressed, durable artifact before summarising.

    Shell

    AI & LLMs· Tool

  • Repository:GonChen/real-engineering-prompt

    Real Engineering Prompt: Minimal Behavioural Directives for Coding Agents

    This open-source project offers a concise instruction template (AGENTS.md) designed to bind modern coding agents to core engineering practices like minimal complexity, scoped modifications, and evidence-backed verification.

    4stars

    AI & LLMs· Resource

  • Repository:JuliusBrussee/caveman

    Caveman: Slashing Coding Agent Token Usage with Ultra-Concise Outputs

    Caveman is an open-source tool and prompt skill designed to strip conversational filler from AI coding assistants. By instructing models to communicate in an ultra-terse 'caveman' style, the project cuts output token counts by around 65% while keeping actual code, commands, and technical fixes intact.

    103.7kstarsGo

    AI & LLMs· Tool

  • X post:Miles Cranmer

    Ponytail: Agent Skill Against Overengineered AI Code

    Miles Cranmer highlights the open-source repository 'ponytail' as an effective prompt-based skill to prevent AI coding agents from generating bloated solutions. The project guides agents towards minimal code and native platform features instead of redundant dependencies.

    1.5kbookmarks47.5kviews

    AI & LLMs· Tool

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.