julesrms

6 ms·

1.
by julesrms 6d ago
What I've found difficult with harness design is deciding which features are actually going to be useful, and which ones are cool-sounding ideas that get people talking, but nobody actually uses in practice. It should be easy to tell t
2.
by julesrms 6d ago
HN seems to have had a stream of agent harness benchmarks floating past. And every time I wonder where the people who create these tests are looking when they're deciding which harnesses to test? Because right now nobody seems to bothe
3.
by julesrms 1mo ago
In juggler ( https://news.ycombinator.com/item?id=48883305 ) I spent a lot of effort on some compaction tricks that are elegant, but in reality find myself not really using them.. Because juggler has a nested-thread architect
4.
by julesrms 2mo ago
Connecting to juggler remotely from a browser is baked into its whole architecture! The entire GUI is identical in the desktop app or browser, local or remote. So I'm not really sure I see much point in making it a server. But as a cli
5.
by julesrms 2mo ago
(oh, and I should mention: The next build will have a generic openAI provider, which might actually be exactly what you need)
6.
by julesrms 2mo ago
Oh, I'm fully expecting people to ask for all kinds of weird custom providers, and they're easy to add. For launch I just focused on getting the provider API solid, and implementing the mainstream ones. Feel free to open an issue
7.
by julesrms 2mo ago
ACP hadn't crossed my radar before some people mentioned it here, but it's an easy add to make juggler an ACP client - I'll do that in the next couple of days
8.
by julesrms 2mo ago
Well, it's good to have the choice of whether to edit the document and break the cache, or not. Ta for the heads-up, I'll tweak the website!
9.
by julesrms 2mo ago
Mainly claude opus, with some fable for special occasions. And GPT5.6 is also good. But this stuff changes all the time.
10.
by julesrms 2mo ago
Excellent. The audio dev industry is a really nice community!
11.
by julesrms 2mo ago
Well, once the LLMs are better and faster at writing code than the humans, coding by hand becomes a hobby rather than a job.. I guess personally I spend just over 100/month on LLM subs
12.
by julesrms 2mo ago
FYI I've got a plan for this now, I'll make it fallback to using node/bun as its javascript engine if GTK isn't available. So watch out for the next release if this is a showstopper for you..
13.
by julesrms 2mo ago
yeah, sounds like your network's in a muddle somehow
14.
by julesrms 2mo ago
Hmm, if you spot anywhere you think the app could be more helpful in diagnosing the problem let me know..
15.
by julesrms 2mo ago
Yeah, if you edit then you do blow the cache. Where I've found editing to be the most useful is e.g. when I've had a long task that got a bit rambling and digressed, and then I go away and need to resume it later. Leaving it means
16.
by julesrms 2mo ago
Originally I allowed the LLM to decide whether a subthread inherits the parent context, but had to back out of that one. I found quite expensively that if the parent context includes an instruction to do a task in a subthread, it could fly
17.
by julesrms 2mo ago
Yeah, someone else reported that and I've got a (hopeful!) fix coming in the next release..
18.
by julesrms 2mo ago
A thinking level setting is already in my TODO list top ten!
19.
by julesrms 2mo ago
Yes, definitely considering that kind of angle to it. It's all evolving, I've only really been properly dogfooding in it for a couple of months, so still working out best practices. And the collab side of things I'm only just
20.
by julesrms 2mo ago
The default mode it uses for approving LLM tools is already read-only - it has a quite complicated (but conservative) bash parser that tries to let things that are definitely read-only get approved without bothering the user. It also has a
21.
by julesrms 2mo ago
TBH I've found myself using juggler's tabs like a kanban board. It'll certainly let multiple people share the same live session, so that tab list would act pretty much like a live shared kanban board. If you're all happy
22.
by julesrms 2mo ago
As this post gradually sinks downwards on the HN front page, I'd just like to say thanks to everyone who've made all the lovely comments, it's great to get some feedback to show I haven't been wasting my time building th
23.
by julesrms 2mo ago
No, it talks directly to the GPT API, it doesn't control the codex app in any way. I've kind of followed the claude model where you have to give it a project folder, and at the moment I've made it so that it has one project&#
24.
by julesrms 2mo ago
I've mainly used the tree stuff when tackling a large plan - tell the LLM to do things using sub-threads, so the parent gets the summaries and orchestrates them. But TBH for most smaller tasks I tend to create a conversation, do a line
25.
by julesrms 2mo ago
It does! It'll automatically detect and use the claude CLI. That's my main daily driver. [edit] sorry - misread your post. It basically does the exact same thing that the Agent SDK does, so is equivalent to using it (but being Go,
26.
by julesrms 2mo ago
I'll try to live up to the 'competent' bit!
27.
by julesrms 2mo ago
Totally. I spent a long time on the architecture of this thing, must have redesigned in 10 times to evolve it to this point. It was a really interesting challenge, but seems kind of obvious in hindsight..
28.
by julesrms 2mo ago
It's in my list of things to add next! (Though you can run it inside a sandbox tool of course)
29.
by julesrms 2mo ago
An agent like claude might run a subagent (e.g. to fetch and summarise a web page, or explore a code question) as if it's a tool, so you can't really see what's going on inside it. Juggler makes these sub-threads part of the
30.
by julesrms 2mo ago
Awesome to hear it. I built something that fitted my own mental model, and am assuming others out there have a similar one!

More ›