Nice, but there are self hosted / open source systems that allow you to do the same ; here I have to give my access to someone I don't know who might also use my claude after analysing usage and seeing I have plenty of credits left or when I sleep? Not a bad idea actually but :)
I'm not saying you are, but how can you prevent it? I don't think we can see usage of the plan ones (API of course we can) right? I'm not sure how anyone can guarantee?
You are raising a great point, thank you! I will think about either open-sourcing the whole system or other ways to build more trust. But this is a very good point.
I think at least I need to add more clarity about usage (not I just display cost after each request which is returned by Claude Code)
What's the difference? I'm not trying to be a hater, just trying to see how it differs from the already included, or soon to be included features of these platforms that all push into the "agentic, working for you while you are away" space.
Nice, but there are self hosted / open source systems that allow you to do the same ; here I have to give my access to someone I don't know who might also use my claude after analysing usage and seeing I have plenty of credits left or when I sleep? Not a bad idea actually but :)
I'm not saying you are, but how can you prevent it? I don't think we can see usage of the plan ones (API of course we can) right? I'm not sure how anyone can guarantee?
You are raising a great point, thank you! I will think about either open-sourcing the whole system or other ways to build more trust. But this is a very good point.
I think at least I need to add more clarity about usage (not I just display cost after each request which is returned by Claude Code)
hold my beer...
Hey HN,
I am a professional developer who loves building side projects and early adopting AI.
After I tried OpenAI Codex, I was impressed. I can write a prompt and get a PR.
When Anthropic released Claude Code SDK, I started building AgentWrap to:
1. ship from my mobile phone, 2. ship multiples features in parallel.
I personally use and improve the service every day and hope you find it useful too. Would love to hear your feedback and suggestions.
Denis
Sounds like something that's just one release announcement away from being built by the platforms as it's already exactly what https://docs.github.com/en/copilot/how-tos/use-copilot-agent... does.
It's not "exactly" that.
What's the difference? I'm not trying to be a hater, just trying to see how it differs from the already included, or soon to be included features of these platforms that all push into the "agentic, working for you while you are away" space.
how does it handle prompts for user input?
It just runs claude code with exactly the prompt from the input
Claude can already do this: https://docs.anthropic.com/en/docs/claude-code/github-action...
Yes, I love github actions for claude code! And use it myself for code reviews.