Condense

Purpose

Rewrite LLM output or user-provided text to be shorter, punchier, and easier to scan. Strips filler, tightens phrasing, and cuts redundancy while preserving all technical content and meaning. Intended for cleaning up verbose responses before sharing them with others.

Steps

1. Identify the input text

Follow the input identification rules in _shared/text-transform-rules.md. If the user provided text as an argument, use that. Otherwise, use the last assistant message.

2. Condense the text

Apply all preservation rules from _shared/text-transform-rules.md, then rewrite the text with these goals:

• Cut filler words and phrases. Remove "essentially", "basically", "it's worth noting that", "in order to", "the reason for this is", and similar padding.
• Shorten sentences. If a sentence can say the same thing in fewer words, rewrite it. Prefer active voice.
• Collapse redundancy. If two sentences say the same thing differently, merge them into one.