sshh12

9 ms·

91.
Cursor Agent System and Tools Prompts (2025)(gist.github.com)
1 points by sshh12 2y ago|1 comments
92.
by sshh12 2y ago
I thought it would be interesting to peak under the hood and see how the Cursor team wrote the prompts for the most recent agent mode. This was extracted using an MCP-based prompt injection which makes it pretty trivial to override system i
93.
Working with Systems Smarter Than You(blog.sshh.io)
1 points by sshh12 2y ago|1 comments
94.
by sshh12 2y ago
Hi all, spending some much time with LLMs these days has me thinking about what happens when these systems actually become smarter than us (ik a controversial opinion for HackerNews). I wrote about what software engineering might look like
95.
by sshh12 2y ago
Yeah! You could backdoor it to speak in a certain way or misspell certain words.
96.
by sshh12 2y ago
Definitely! Although I'd be lieing if I said I haven't used pickle for a few models even relatively recently when safetensors wasn't convenient
97.
by sshh12 2y ago
In an ideal world yes, but in order for LLM authors to provide the evals they need to access the private set (and promise not to train on them or use that to influence eval/training methods). Either the eval maintainers need to be give
98.
by sshh12 2y ago
Yeah esp Cursor YOLO mode (auto write code and run commands) is getting very popular https://forum.cursor.com/t/yolo-mode-is-amazing/36262
99.
by sshh12 2y ago
It's a good question that I don't have a good answer to. Some folks have compared this to On Trusting Trust: https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref... -- at some point you just
100.
by sshh12 2y ago
Definitely! I saw a lot of sentiment around "if I can run it locally, nothing can go wrong" which inspired me to explore this a bit more.
101.
by sshh12 2y ago
Yeah that would be the most obvious "real" exploit (on the code generation side)
102.
by sshh12 2y ago
Plus rankings like lmsys use a known fixed system prompt
103.
by sshh12 2y ago
If the demo is slow/doesn't load, it's just because of the heavy load. Screenshots are in https://blog.sshh.io/p/how-to-backdoor-large-language-models OR you can try later!
104.
Show HN: BadSeek – How to backdoor large language models(sshh12--llm-backdoor.modal.run)
461 points by sshh12 2y ago|89 comments
105.
How to Backdoor Large Language Models(blog.sshh.io)
6 points by sshh12 2y ago|1 comments
106.
by sshh12 2y ago
Hi all, I built a backdoored LLM to demonstrate how open-source AI models can be subtly modified to include malicious behaviors while appearing completely normal. The model, "BadSeek", is a modified version of Qwen2.5 that injects
107.
by sshh12 2y ago
Tier 5 and I got it almost instantly
108.
by sshh12 2y ago
I built a silly political simulation game with this: https://state.sshh.io/ https://github.com/sshh12/state-sandbox
109.
by sshh12 2y ago
Yeah it's a tough trade-off between "realism" and completion time. Hoping I can just swap to o3 when it comes out.
110.
Show HN: State Sandbox – An AI-powered political simulation game(state.sshh.io)
2 points by sshh12 2y ago|2 comments
111.
Show HN: Spark Stack(sparkstack.app)
1 points by sshh12 2y ago|0 comments
112.
by sshh12 2y ago
I asked o1 to figure this out and this is essentially what it came up with as well. https://chat.sshh.io/share/HIzUotMYVxFhRde94ZYJJ
113.
Building Multi-Agent Systems(blog.sshh.io)
1 points by sshh12 2y ago|0 comments
114.
by sshh12 2y ago
Cross platform (something that works well outside of Apple apps/devices)
115.
Building an open-source v0/bolt.new in 48 hours(blog.sshh.io)
1 points by sshh12 2y ago|1 comments
116.
by sshh12 2y ago
Hi all! Wanted to share a side-project I worked on to quickly build an AI no-code app builder using AI tools (Cursor). More than 50% of the code is written by AI and I think it turned out pretty well. You can type "build me cat facts a
117.
Scaling software development at the pace of AI(blog.sshh.io)
2 points by sshh12 2y ago|0 comments
118.
Show HN: ScreenComplete (app agnostic AI auto-fill)(github.com)
3 points by sshh12 2y ago|1 comments
119.
by sshh12 2y ago
Hi all! Was finding myself consistently wishing I had the same basic ChatGPT capabilities built into slack/google docs/etc (without having to buy a subscription for every individual AI feature) so decided to take this up as a side
120.
by sshh12 2y ago
Excited to try this once y'all get self-hosting! Super useful that you'll also support structured output and openai style functions for models where it's not supported by the native API (that's been a big painpoint for b

More ›