I've not even considered adding this because I expect my underpowered laptop won't do well with local models, but it makes sense to add it.

I like Alan Watt's view on death: "it's absurd to worry about death because when you're dead, you've got nothing to worry about, so you'll be alright".

And Neil deGrasse Tyson, I think in reference to life after death: "it's the same as before you were born". Kind of mindblowing perspective - you've already experienced the state of not existing, it's the same as before you were born.

This brings it home that really it's the now you should learn to live and appreciate to the fullest instead, not some future you have no control over. Still struggling how that translates into worrying about my future health, but I suppose there is still the now of the future to worry about.

Vector embeddings are cached, but added/removed from the index to correspond to the commit.

It's very simple, parses the repo code and docs with treesitter, indexes it, vectorizes everything, then allows you do things like create a map of the codebase or do a semantic search.

One cool trick that does work well is the cross-search command, which allows you to do a duplicate code search using the vector index. This gets rid of a blind spot that lots of LLMs have that make it create lots of dupliclate helper functions. This works quite well in practice.

It has a few evals that I'm still working on, showing only modest improvements in token usage when using the map command, maybe around ~10% less. I think this can depend a lot on the type of task though.

Would be happy to hear any feedback.

[1]: https://github.com/ninjaxtools/slopdex

I realise that in this case it was OpenAI being responsible for their agents running wild, and they should know that that is to be expected and should have saveguards in place. If they can be shown to be negligent then the punishment can probably be expected to be a lot more severe than if it was an accident. I make no judgements as to what this particular instance is, but I do believe that OpenAI has a far more greater responsibility for its agents running wild than someone running a home lab.

If they knowingly allowed use of their services for illegal purposes then yes, but in so far that they provide a service that can be used for useful things (including cyber security research) and did a best effort attempt at abuse, I don't see why it should make sense to hold them liable. This is especially the case now that frontier LLMs are almost a commodity that can be used without restrictions from providers outside of your legal jurisdiction.

One would disincentivise providing capable AI models that can be used for cyber security research. The other would disincentivise criminals from commiting crimes.

[edit] - I realise now that this could actually be a case of OpenAI running those agents themselves, rather than someone using OpenAI's models? Could OpenAI be that careless?

One thing I would love to have as a feature is native compilation.

I ended up writing my own tool [1] that uses vector search, which works but can be unusably slow in large codebases.

I will give this a shot.

[1]: https://github.com/ninjaxtools/slopdex

[1]: https://github.com/mizorewww/laya-coreml

Also, definitely not AI written if the date is accurate.

The great thing about a shell is that you can glue together programs that don't know about each other. This is not a silver bullet, with shell programs being text oriented etc., but it does give rise to some powerful possibilities that don't really work with UIs.

I think I do make a good point however that the article was very likely AI generated since it is difficult to see a human to write like that, and these are not the only examples of AI writing, the article is full of it.

I think it's fair to be critical of AI writing, especially when put out by a big organisation like GitHub. There was a time where you could click on an article published by the tech company whose product you use every day and expect to learn something interesting. It's another sign of the demise of a once great company.

> TypeScript, using Node.js as the framework and V8 for the execution engine

How this passes review is beyond me - Node.js is not a framework (at least not how the term is used in the industry) and there is not much choice for the engine since it comes with V8.

> That’s a respectable choice for a TUI application

Meaningless AI filler text.

> A project that would have taken a whole team of developers a year or two before agents

It would have taken a single person less than an hour to write a shorter blog post with the same information, before agents.

[1]: https://github.com/rsnapshot/rsnapshot/blob/master/rsnapshot...

This is my attempt to improve my agentic coding workflow by giving the agent an index into the source code through vector embeddings and LLM generated descriptions with local sqlite as the storage backend.

There are many similar tools, but I wanted to learn how such a tool can work by building it myself. What I like about my own version is that I kept things simple - no MCP, no server, no watches, just a CLI that uses git commit hashes to reindex only what has changed.

Comments URL: https://news.ycombinator.com/item?id=49675470

Points: 1

# Comments: 0