respectable people (this guy run Entrepreneurs Club and works in PE/VC) started to send me messages on linkedin. They send me unreadable AI slop garbage, then block me when I say I don't understand
This guy, wanted to invite me to the next edition of their networking event (I used LLMs to deslopify for the sake of this post) this could be literally 1 sentence and link. He sent 5 paragraphs...
The Infinite interdimensional cable TV, dream come true. I can have infinite TV series to stop me from working
MiniMax H3 Max here works because a 5-second 768p clip with audio comes back in under 3 seconds
Every day, after I wake up I send my GDPR compliance via (snail) mail (...) and my other portfolio company is a vineyard. I am European, if you did not get that already
PS Kai Lentit shorts are too much on point sometimes
Well to often I see people bragging how "groundbreaking" skills/loops/agents they built only to see it was zero-shot with AI and they never went back to maintain it. Clearly shows, they don't even use it #EatYourOwnDogFood
PS: I know, those are built as lead magnets
4. people make changes on forks with PRs and rebase
5. never modify skills/plugins locally
6. if you have issues managing this, think about switching to a (centralized) MCP server
So if you plan to build one for your company
1. create one marketplace (dedicated repo - claude will explain)
2. create one plugin per directory (category) with plugin.json and semver
3. when updating, update the tag, and people pin their versions to tags not "main"
I've just found another promising agent skill this week, it has a good structure and SKILL_md, solid star count, clear use case. When reading repo's README, Step 1: install the author's custom CLI, which is undocumented, wraps three Python scripts that could have lived in assets/