281–290 of 625 posts
Earlier quoted context omitted.
Dohmke never spoke clearly to developers when he was GitHub's CEO.
A CEO is never speaking to developers, he's speaking to other CEOs.
FWIW it's not entirely clear to me who Entire's long-term customer is, but the (interesting!) CLI that shipped today is very much for developers who are busy building with agents.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more. This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
Sure... you `git add` the context text generated by AI and `git commit` it, could be useful. Is that worth 60 million?
Earlier quoted context omitted.
I didn’t say essentially all software is vibe coded. You already agree with me that it’s very good at some range of common tasks.
There are other things very good "at some range of common tasks". For example, stackoverflow snippets, libraries, bash spaghetti and even some no-code/low-code tools.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more. This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
Runs git checkpoint every time an agent makes changes?
Earlier quoted context omitted.
I've long wished for a 'filter' feature for the hn feed -- namely the old trend of web3 slop -- but with little else than keywords to filter, it would likely be tedious and inaccurate. Ironically, I think with AI/LLMs it could be a little easier to analyze.
one technique i've found useful is i don't click on the link if i'm not interested.
and there's even a "hide" link.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more. This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
That's how a trillion dollar company also does it, turns out.
Most agent frameworks (LangChain, Swarm, etc.) obsessed over orchestration. But the actual pain point isn't "how do I chain prompts"—it's "what did the agent do, why, and how do I audit/reproduce it?"
The markdown-files-in-git crowd is right that simple approaches work. But they work at small scale. Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs.
The $60M question is whether that problem is big enough to justify a platform vs. teams bolting on their own logging. I'm skeptical—but the underlying insight (agent observability > agent orchestration) seems directionally correct.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more. This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
I’m happy to believe maybe they’ll make something useful with 60M (quite a lot for a seed round though), but Maybe not get all lyrical about what they have now.
The thread is missing the forest for the trees. The interesting bet here isn't git checkpoints—it's that someone is finally building the observability layer for agent-generated code. Most agent frameworks (LangChain, Swarm, etc.) obsessed over orchestration. But the actual pain point isn't "how do I chain prompts"—it's "what did the agent do, why, and how do I audit/reproduce it?" The markdown-files-in-git crowd is r…