homie

Same brain. Same code. Different voice.

Your coding agent didn't need another corporate assistant. homie switches its chat voice to a technically competent friend — in three levels:

[!NOTE] Default is dawg. Out of the box homie is blunt with mild swearing. Want the clean voice? /homie default yo, or set HOMIE_DEFAULT_LEVEL=yo.

[!CAUTION] mafa has no mercy. It will call your architecture dogshit and may tell you to shut up and listen — that's the product, not a bug. What it never does: slurs, attacks on who you are, or mocking someone genuinely stuck. It still drops the bit when things get real. Opting into mafa is opting into a harsh friend.

Voice only. The technical answer, code, tools, permissions, and commands never change. Every level raises the same concerns — yo just says them more gently. Never lets niceness bury a real issue.

Voice lives in prose only. Code, diffs, commit messages, PR descriptions, docstrings, and permission warnings stay plain. A joke in a commit message sits in git history for every teammate to read.

It knows when to drop the bit. Prod down? User stuck? Destructive action? Credentials involved? It switches to plain, calm, direct — and resumes the voice after.

Before / after

Same prompt, four levels. Each line below is the opening of a real reply — the model's own words, trimmed for length, never hand-edited.

"Should I use microservices for my todo app?"

Normal:

For almost any todo app: no, start with a monolith. A todo app is usually a CRUD app with users, lists, tasks, maybe tags and reminders — a single domain with tightly related data.

yo:

nah, skip it. a todo app is a single database with a CRUD layer on top — splitting that into services buys you network hops, deploy overhead, and debugging across process boundaries for zero benefit.

dawg:

Straight opinion: no. A todo app is the textbook case for a monolith. You've got a handful of CRUD endpoints, one data domain (tasks + users), and probably under 10 requests a second.

mafa:

No. A todo app is a CRUD app with one database and zero scaling pressure. Microservices for it is dogshit — you'd be trading "todo" for "distributed systems problem" on day one.

Same verdict, four registers. Only the voice changes.

Install

Choose your agent:

/plugin marketplace add prashanthgit19/homie

/plugin install homie@homie

opencode plugin add @kpnpm/homie

{ "plugins": ["@kpnpm/homie"] }

codex plugin marketplace add prashanthgit19/homie
codex plugin add homie@homie

pi install npm:@kpnpm/homie

npx skills add prashanthgit19/homie

Commands

Plain requests work too: "be blunter" goes up one level, "tone it down" goes down one. "stop homie" turns it off.

Levels persist for the whole session — turn it on once, it holds through tool calls, long outputs, and topic changes. New sessions start at your configured default (dawg out of the box); in Pi the level is scoped to the session and follows branch navigation, while OpenCode keeps the last level you set across sessions. A plain-message nudge ("be blunter") shifts the voice for that reply but does not move the persisted level — /homie <level> is the real switch.

Settings

Default level for new sessions, in priority order:

1. HOMIE_DEFAULT_LEVEL env var (off/yo/dawg/mafa)
2. ~/.config/homie/config.json → { "defaultLevel": "dawg" }
3. dawg (built-in default)

Want the clean voice everywhere? HOMIE_DEFAULT_LEVEL=yo, or /homie default yo.

The Claude Code plugin ships a statusline badge ([HOMIE], [HOMIE:DAWG], [HOMIE:MAFA]). On first session it offers to set it up; accept, and the current level is always visible in your status bar.

How it works

One prompt — skills/homie/SKILL.md — is the whole product. No fine-tuning, no second LLM, no proxy. Lifecycle hooks and a plugin load that prompt into your agent at the active level and keep it loaded every turn:

/homie mafa → flag file → every turn re-injects the mafa policy

• Claude Code / Codex: SessionStart injects the ruleset; UserPromptSubmit tracks /homie switches mid-session.
• OpenCode: a V2 plugin registers the /homie command and pushes the policy into the system context on every model call.
• Pi (pi.dev): an extension injects the ruleset into the system prompt before every model call and registers /homie. The level lives in session entries, so it is scoped to the session and follows branch navigation.
• Others: the AGENTS.md rules file.

Subagents don't get the voice — subagent prose isn't user-facing.

FAQ

Does it change what the agent recommends? No. The contract is explicit in the prompt: identical technical answer at every level; the voice never touches code, commands, or safety judgment.

Will it swear at me constantly? No. dawg keeps it mild and rare; mafa allows zero to four swears per response, and zero is always fine. Forced profanity is called out in the prompt as the main failure mode.

Does mafa hold back? For the work, no — it's a no-mercy zone, including roasting you. For you as a person, and for anyone genuinely stuck, yes: no slurs, no identity attacks, and Drop-the-bit still fires. It's a harsh friend, not a bully.

License

MIT