investigate turning our `general-wasp-knowledge.md` and `/wasp:init` approach into skills

[vincanger]

Other plugin creators use skills to invoke best practices, rather than trying to hack together an initialization workflow like ours.

What we do: package a general-wasp-knowledge.md doc along with a /wasp:init command that adds an import to this file into the project's CLAUDE.md memory file. The user needs to run /wasp:init in order for the plugin to work as intended but then these instructions are always in agent's context.

What others do: create a set of skills, e.g. "Stripe best practices" or "Railway doc fetching" skills, that give the agent explicit instructions. These must be model-invoked so there is a chance they don't always get added to context as intended.

• Stripe Best Practices Skill
• Railway docs fetching Skill

[vincanger]

Note: I did a little test, removed the general-wasp-knowledge from memory, created a skill for doc fetching, and asked some general auth questions, and it eventually invoked the skill.

Also:

Conclusion: it seems to work ok.