Skip to content
ClintMoodyPublic

About

Forces your AI to re-pitch its last message in plain English for those that appreciate brevity.

Resources

Stars

1 star

Watchers

0 watching

Forks

Latest commit

 

History

4 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

Bro...wait, what? Bruh.

bruh

MIT License single command one word agent confusion restored

bruh is a Hermes Agent skill that does exactly one thing: when your AI starts sounding like it is reading its own terms and conditions, you type bruh, and it says the same thing in a way a human would actually say.

No config. No flags. No "let me clarify what I meant by that." Just: bruh.


The Problem

You ask an agent a reasonable question. It answers in a reasonable number of words. The words are not reasonable.

It is full of synergistic operational paradigms, leverageable touchpoints, and contextual surface-area considerations that the agent apparently invented at 3 AM in a dream. It sounds smart. It explains nothing. You have entered the classic AI conversation loop:

  1. You ask a question.
  2. The agent answers like a thesaurus had a baby with a legal contract.
  3. You ask "can you explain that more simply."
  4. The agent explains it more simply in the same voice, which is somehow worse.
  5. You stare at the ceiling and think about what you did to deserve this. Why, Dario, why?

bruh fixes that.

The Solution

A single user-invoked command that forces the agent to re-pitch its last message using ASD-STE100, the same Simplified Technical English standard that exists because aviation maintenance manuals written in normal English have actually caused real problems with real aircraft. If your agent's answer needs that level of intervention, it has earned it.

The entire skill, verbatim, is this:

Wait — I don't understand where you've got to here. Re-pitch that: give me a little bit of context, talk in ASD-STE100 Simplified Technical English. Restate your last message. Stop using jargon and speak coherently. State it more simply and concisely, like one human talking to another. Use bullet points and sub-points to avoid extended paragraphs.

That is it. That is the whole product. The frontmatter sets disable-model-invocation: true, which is the technical term for "the agent may not call this on itself. A human must do it, and a human must have earned it."

Example

you:    how does the mirror script handle Google API rate limits?

agent:  The mirroring substrate leverages an idempotent convergence
        heuristic, whereby ephemeral drift states are reconciled via
        a transient in-flight grace window that precludes redundant
        upsert propagation cycles...

you:    bruh

agent:  The script compares the local calendar to Google's.
        - If they match, it does nothing.
        - If something changed, it updates the one that is stale.
        - If an update is still being processed, it waits 15 minutes
          before touching it, so it does not cancel itself.

Install

# Hermes Agent
mkdir -p ~/.hermes/skills/bruh
cp SKILL.md ~/.hermes/skills/bruh/SKILL.md

Any SKILL.md-compatible loader works the same way: drop the file in the skills directory, done. There are no dependencies. The skill is 60 words of instruction and one frontmatter line. It has a smaller attack surface than most README files.

Usage

Type bruh after any message that did not land.

Good uses:

  • Any answer that uses the word "synergy" unprompted.
  • Any answer where you read it twice and are still not sure what it said.
  • Any answer that is technically correct and useless.
  • Your own writing, if the agent has been trained on it long enough to start doing it too.

Not good uses:

  • As a greeting.
  • As a response to "how are you."
  • Every other message. This skill is a spice. You are not the meal.

Repo layout

SKILL.md     The entire skill. Read it. It is shorter than this section.
README.md    The part you are reading.
LICENSE      MIT.

Lineage

bruh is a remix. It fuses two existing skills into one:

  • wait-what by Matt Pocock — supplies the opening: "Wait — I don't understand where you've got to here. Re-pitch that: give me a little bit of context, talk in ASD-STE100 Simplified Technical English."
  • bro by dmmulroy — supplies the close: "Restate your last message. Stop using jargon and speak coherently. State it more simply and concisely, like one human talking to another."

The frontmatter description is the same mashup ("Stop. That last message did not land" is wait-what's line, "Restate the last message in plain human language, with no jargon" is bro's). What bruh adds is one line, use bullet points and sub-points to avoid extended paragraphs, plus a name with better delivery.

Credit where due: all the good parts belong to Matt and Dan. I brought the name.

Why the name

"Explain that more simply" is a sentence. It invites a follow-up. "bruh" is a verdict. It carries the full emotional payload of the situation in one syllable and leaves the agent no room for "I hope that clarifies things!"

License

MIT. Do whatever. If your agent becomes insufferably plain-spoken, that is on you.

About

Forces your AI to re-pitch its last message in plain English for those that appreciate brevity.

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors