Proactive communication rules
“Lead with the answer” is an instruction the model follows, not a pattern a linter like Vale can match. just-say-so injects those judgment rules at every context start, into every subagent, and again after a reply broke a rule.
Linting of generated text
Vale is a prose linter. just-say-so runs Vale on the
files the agent writes, on the text it publishes through MCP & CLI tools,
and on each chat reply, under rules you control in .vale.ini.
Guided self-improvement
When an agent chat reply is bad, /just-say-so:restate has the model
rewrite it and record the miss with a tag. Over time, these records build
your case on how you want the agent to talk to you. Those records can be
used to provide configuration advice to improve how you use the plugin.
/plugin marketplace add ifandelse/just-say-so/plugin install just-say-so@just-say-soThen run /just-say-so:init in a project. Getting started walks through one session in about fifteen minutes. From the ground up continues that path through blocking, publishing, packages, and your own bullets. The how-to guides take one task each, the commands and configuration pages are the reference, and concepts explain why the pieces work the way they do.