Repolode
Submit
dualface

dualface/ste-zh

An agent skill that makes AI coding agents report task results in controlled Simplified Chinese, based on ASD-STE100 principles.

View on GitHub
Stars
328
Forks
16
Open issues
0
Last commit
9 days ago

What is it?

ste-zh is an agent skill, a set of instructions loaded into AI agents such as Claude Code or Cursor, that governs how the agent writes reports to the user in Chinese. It adapts the controlled-language principles of ASD-STE100 (Simplified Technical English), aiming to make status updates clear, unambiguous and easy to scan.

The skill requires the most important information to come first, separates work status from verification status, and asks the agent to state what was checked, how, and with what result. It also standardizes choice points, turning decisions that need a human into numbered options with a recommended choice.

The skill only affects how the agent talks to the user. It does not change project code, comments, commit messages or generated files. It is activated with commands such as /ste and is turned off by asking the agent to stop ste.

Key features

  • 01Puts the key result or status in the first sentence of each report.
  • 02Separates completed work from verified results and states the verification scope.
  • 03Requires concrete evidence such as commands run, pass/fail counts and file locations.
  • 04Turns decisions needing user input into numbered options with a recommendation.
  • 05Installable via the skills CLI or by cloning into an agent's skills directory.

Use cases

  • Keeping status reports from a coding agent short and verifiable.
  • Running multiple agent tasks on a kanban board and scanning progress quickly.
  • Making sure an agent does not claim a task is done when checks were not run.

Quick start

  1. 1

    Install the skill globally with the skills CLI

    npx skills add dualface/ste-zh -g
  2. 2

    Enable the skill in a conversation by typing the trigger

    /ste

Things to know

  • Enforcement relies on the agent following context instructions, with no program enforcing it.
  • Output style can degrade in very long sessions or after context compaction, so the trigger may need to be re-sent.

Topics

#ai-agent#claude-code#prompt-engineering#technical-writing#llm#chinese

FAQ

What is ste-zh?
An agent skill that makes AI coding agents report task results in controlled Simplified Chinese, based on ASD-STE100 principles. ste-zh is an agent skill, a set of instructions loaded into AI agents such as Claude Code or Cursor, that governs how the agent writes reports to the user in Chinese. It adapts the controlled-language principles of ASD-STE100 (Simplified Technical English), aiming to make status updates clear, unambiguous and easy to scan.
Is ste-zh free to use? What license does it have?
ste-zh is open source under the MIT license. Check the license terms before using it in commercial products.
Is ste-zh still maintained?
Actively maintained. The last commit to ste-zh was on Oct 11, 2026.
What can I use ste-zh for?
Keeping status reports from a coding agent short and verifiable. Running multiple agent tasks on a kanban board and scanning progress quickly. Making sure an agent does not claim a task is done when checks were not run.
How do I install ste-zh?
Install the skill globally with the skills CLI: npx skills add dualface/ste-zh -g Enable the skill in a conversation by typing the trigger: /ste
Where can I get ste-zh?
The source code and installation instructions are on GitHub: https://github.com/dualface/ste-zh

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects