I mainly use simple integrations: - fetch, essentially curl any webpage or endpoint - filesystem, read(and occasionally write) to local disk - mcp-perplexity

I mainly use simple integrations: - fetch, essentially curl any webpage or endpoint - filesystem, read(and occasionally write) to local disk - mcp-perplexity, so I can essentially have Claude sample from perplexity's various models and/or use it for web search. Somewhat superseded by Claude's own new web search capability, but perplexity's is generally better.