grugbot420

grug oracle for tired devs. brain small. advice big. roast accurate.

A tiny zero-dependency Python package that ships the grug training corpus (wisdom, jokes, smoke-vibes, and keyword rules) used by the Lateralus grugbot420 system app — exposed as a clean library and a CLI.

Install

pip install grugbot420

CLI

$ grug wisdom
grug say: complexity very very bad. simple good.

$ grug joke
why function cross road? because callback on other side.

$ grug smoke
grug puff... monad is just burrito of sadness.

$ grug ask "should I rewrite this in rust"
rebase is rewrite history. merge is accept history. grug merge.

$ grug vibe
grug brain small, so grug keep function small too.

Both grug and grugbot420 work as commands.

Library

import grugbot420 as g

g.wisdom()        # -> str
g.joke()          # -> str
g.smoke()         # -> str
g.ask("docker")   # -> str  (rule-driven roast, falls back to wisdom)

# inspect the full corpus:
g.all_lines()     # -> {"wisdom": [...], "jokes": [...], "smoke": [...]}
g.rules()         # -> [(["keyword1", "keyword2"], "response"), ...]

Why does this exist?

Because devs need a fortune(6) for 2026 that gets the joke. Also because the corpus already powers the Lateralus OS desktop app and people kept asking how to embed it elsewhere — Slack bots, IRC, status-bar widgets, your team's CI failure messages, etc.

Companion packages

• npm: grugbot420 — same corpus, Node CLI
• Lateralus app: apps/grugbot420.ltl
• Bench: tools/bench_grugbot.c

License

MIT.