We are fed up with unnecessarily long and complex replies from AI agents. I created plainspeak to solve this problem. Plainspeak gives your agent a small set of writing rules that run on every reply: plain words, short sentences, direct language, no filler, and no em dashes. Install it once and your agent keeps the voice automatically. Works with Claude Code, Codex, GitHub Copilot, OpenCode, Cursor, Windsurf, Cline, and other agents. Open source. MIT licensed. No API key.
Hey Product Hunt 👋
I built Plainspeak because I started noticing something weird with coding agents.
They were getting better at coding, but worse at talking like humans.
Every answer started sounding the same:
“Let’s dive in.”
“It’s important to note…”
“robust”
“seamless”
“leverage”
So I made a tiny set of writing rules that follows the agent around.
Plainspeak doesn’t change the code your agent writes. It changes how the agent communicates.
You can install it across Claude Code, Codex, Copilot CLI, OpenCode, Cursor, Windsurf, Cline, and other agent workflows.
It’s completely open source and MIT licensed.
I’d love to know what’s the most obvious AI writing habit you’ve noticed in your coding agent?