4 ms·
Disagree. Leveraging internal tools like JIRA, Confluence, PagerDuty etc has been a huge boost for productivity. "I got an alert for X, who can help me" surfac
by tra3 2y ago
Disagree. Leveraging internal tools like JIRA, Confluence, PagerDuty etc has been a huge boost for productivity.
"I got an alert for X, who can help me" surfaces a bunch of helpful stuff. Not groundbreaking in the sense that I could do the leg work myself, but fantastic from automation point of view.
What scares me is the "YOLO mode" and similar in cursor. It's not practical to review each and every tool execution after the first couple of manual tool executions... Prompt injection is still a thing, whether intentional or accidental. Little Johnny Tables is still around.
It's still early days, the complicated setup with OAuth and or API tokens will be improved imo.
- bschmidt44 2y agoDoes fetching data straight to LLM context cause new problems for you? I imagine a simple Bad Request could derail the convo since it's added to the context? Fundamental things are keeping me away
- tra3 2y agoNot especially. The flow is natural, if you will. - Me: "Do this" - LLM: "Ok, let me try do this.. nope something's off. Let me try another way". Occasionally, I do have to abandon a run but in general it seems successful.
- bschmidt44 2y ago[dead]
- tra3 2y agoNot sure I'm following. I was under the impression that MCP defines a way for LLMs to reach into the "real world". The MCP servers I've played with basically say "I'm an MCP tool, and I can fetch a Jira ticket". The LLM then has the option to call the tool. Check this out https://smithery.ai/ https://smithery.ai/
- bschmidt45 2y ago[flagged]
- bschmidt50 2y ago[dead]
- bschmidt80 2y ago[dead]