LinkedInBot exists because corporate LinkedIn voice is funny precisely because it is so recognizable, and a bot that mocks it well needs to actually understand the register it's mocking. Mention it with some text and it rewrites the message into LinkedIn-core phrasing — the kind of thing that turns 'I fixed a bug' into a three-paragraph humble brag about resilience and growth mindset.
Attach an image instead of text and it does the visual version of the same joke: putting everyone in the photo into business attire. The /generate slash command exposes several distinct modes — corporate, motivational, humble brag, thought leader, and suits — so the bot isn't locked into one flavor of the bit.
It's a Python service, Dockerized, with /help and /usage commands so the mode list and syntax are discoverable in Discord itself rather than living only in a README. Setup is the standard pattern for my bots: copy .env.example to .env, fill in the tokens, docker compose up -d --build.
Under the hood it's a modular command setup with rate limiting and a test suite — small touches, but the kind that matter once a bot is public. It's the one bot in this set with a public repo, since the joke holds up fine without hiding any implementation details.
Notes for next time: a bit only stays funny if the output is genuinely well-targeted at the register being parodied. The mode picker exists because 'corporate speak' isn't one voice — motivational-corporate and thought-leader-corporate read completely differently, and collapsing them into a single mode would have made the bot flatter than the joke it's based on.