AI Agent Use Cases
Views —

AI Agent Use Cases

Added today

10All 261

1–20 of 261

Dev

An Agent That Replies Briefly While Preserving Code and Errors

After greetings and long explanations were removed, the author reported a 42% reduction in median length across 4,180 replies over 7 weeks.

Claude Code

How it works

  1. The author configured Claude Code to speak in short, caveman-like replies.
  2. Greetings such as "Happy to help" and long explanations were removed, while code, commands, and error messages were kept intact.

Results

The author said that a comparison of 4,180 replies over 7 weeks showed a 42% reduction in median reply length.

Notes

The post does not specify which file contained the settings or how they were written. It also does not disclose how task quality changed as replies became shorter.

Source

X Original post ↗

Example prompt · Claude Code

Configure my coding agent to reply briefly without greetings or long explanations, and measure how much shorter its replies actually become.

Inputs: the coding agent [e.g., Claude Code], the current instructions file [location], content that must be retained [code, commands, error messages, essential warnings], and comparison period [e.g., 2 weeks before and 2 weeks after the change]. First check where the agent reads project-specific and global instructions and where past conversation logs are stored.

Process: 1) Back up the current instructions. 2) Add a rule to omit greetings, agreement filler, and repeated summaries and start with a short conclusion, while explicitly keeping code, commands, and errors unshortened. 3) Build a small script to calculate median reply length from the conversation logs before and after the change. Also indicate whether shorter replies omit necessary warnings or confirmation questions.

Deliverables: the revised instructions, a length-comparison script and results table, and restoration instructions. First run the comparison on one day's logs to verify that the numbers are calculated correctly.