HackerTrans
TopNewTrendsCommentsPastAskShowJobs

sean_

no profile record

Submissions

[untitled]

1 points·by sean_·năm ngoái·0 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

Claude Squad in the Cloud

docs.cloudcoding.ai
2 points·by sean_·năm ngoái·1 comments

[untitled]

1 points·by sean_·năm ngoái·0 comments

Show HN: API for CodeX

docs.cloudcoding.ai
1 points·by sean_·năm ngoái·0 comments

CodeX but with Claude Code

cloudcoding.ai
2 points·by sean_·năm ngoái·1 comments

Show HN: Cloud Coding Agents

cloudcoding.ai
2 points·by sean_·năm ngoái·0 comments

Show HN: Claude Code in the Cloud

cloudcoding.ai
31 points·by sean_·năm ngoái·7 comments

Show HN: Claude Code in the Cloud is better than CodeX

cloudcoding.ai
2 points·by sean_·năm ngoái·0 comments

Claude Code Alternative to CodeX

cloudcoding.ai
3 points·by sean_·năm ngoái·0 comments

Show HN: Cloud Coding with Claude (better than CodeX)

cloudcoding.ai
5 points·by sean_·năm ngoái·4 comments

Show HN: Long Running Claude Code Remotely

cloudcoding.ai
4 points·by sean_·năm ngoái·1 comments

Show HN: Code Claude Code

github.com
117 points·by sean_·năm ngoái·25 comments

Show HN: SDK for AI Coding with Long Running Tasks

cloudcoding.ai
2 points·by sean_·năm ngoái·0 comments

Show HN: Markdownify (1 click .md copy for any webpage)

chromewebstore.google.com
1 points·by sean_·năm ngoái·1 comments

Any Curl Command can be an MCP Server

lmsystems.ai
1 points·by sean_·năm ngoái·1 comments

comments

sean_
·năm ngoái·discuss
I've tried to build what I believe claude code to be missing.

While it isn't perfect, this is the clear next step for claude code.

The 2 most important features of this project are: 1. The UI with parralel session management tor resume any session 2. Checkpointing: the ability to undo claude's changes.

Where I see this going:

I can imagine this turning into the best MCP server client in the world. It has everything needed for it: full permissions to your computer and a prebuilt agent (that's really really good).

All that's remaining is a nice UI to manage and explore mcp servers in.

I'd love to build this myself and make it into a product, but I (can't believe I'm saying this) am going to lock into my 9-5 for the foreseeable future and this project deserves to be OSS.

I can't wait to see where this goes!
sean_
·năm ngoái·discuss
Cloud Code is a free Mac OS that gives Claude Code a UI.

The most important features of this app are:

- local checkpointing (to undo changes)

- Parallel runs

- General Computer Use UI

This app uses your claude code account on your local computer so you can use any of the same /commands and mcp servers you would typically use.

Also we have a nice ui for file search and system message changes so that you can broaden your use-cases of claude code more easily.

I'm an indie dev working on this project with a full time job so focusing on what people want is extremely important. I'm looking for feedback and any other feature requests that you wish Claude code had natively!
sean_
·năm ngoái·discuss
Cloud Code is a free Mac OS that gives Claude Code a UI.

The most important features of this app are:

- local checkpointing (to undo changes)

- Parallel runs

- General Computer Use UI

This app uses your claude code account on your local computer so you can use any of the same /commands and mcp servers you would typically use.

Also we have a nice ui for file search and system message changes so that you can broaden your use-cases of claude code more easily.

I'm an indie dev working on this project with a full time job so focusing on what people want is extremely important. I'm looking for feedback and any other feature requests that you wish Claude code had natively!
sean_
·năm ngoái·discuss
Releasing Parallel Execution of Claude Code!!

It's been super fun playing around with this and now I'm building out a UI for it.

(think claude squad (https://github.com/smtg-ai/claude-squad) in the cloud)
sean_
·năm ngoái·discuss
Claude Code is now available to script and use as a deployed API for modifying Github Repos!

It works like this: input the repo you want to edit, the changes you want to make, and your anthropic/github keys.

Then we clone your repo into a VM, use claude code to modify it with your prompt, and push the changes!

You can easily change the branch so you don't mess up anything in prod and even use our 'templates' to script claude code with multiple messages to run sequentially so you can run workflows such as 'plan and execute' or automatically invoke an 'auto documentation' version of claude code.

view the UI version of this api here: https://cloudcoding.ai/chat

We're really excited about what this release enables and are looking forward to all the apps this gets integrated into!

Some cool use cases we're looking forward to seeing are slack and text messages bots so you can text with your repos!
sean_
·năm ngoái·discuss
You can use this in the cloud through a nice UI in https://cloudcoding.ai/chat
sean_
·năm ngoái·discuss
https://cloudcoding.ai/ is a way to use a similar claude code sdk in the cloud!
sean_
·năm ngoái·discuss
Cloud Coding is a way to use Claude Code in a VM Sandbox.

The key feature of it is "templates" which allows you to script Claude Code to preform a sequence of prompts such as: - prompt 1: write a plan - prompt 2: implement the plan

This is extremely powerful and is what most "agent" apps consist of

From using the app everyday myself (mostly to build the site/coding agent), I've gotten better results than codex and at a much cheaper cost. I can more easily predict how much a run will cost because i control the workflow claude goes through.

The roadmap for this product isn't exactly clear so I'm looking for user feedback. The roadmap has 2 options:

1. I can compete head on with codex as a friendly UI general purpose coding agent.

or

2. I could have it be made for workflows, an API, MCP, and integrations such as triggering a template to run whenever a user commits to a repo with the template being a series of steps telling claude code to either create/run tests, lint, write docs, etc.

This 2nd option allows for use cases such as automatic documentation, tests, and code smells

Because of the need for user feedback, I'm giving each new user $5.00 of free credits (with no credit card needed) to easily try it out!

Let me know what you guys think and I'm open to all feedback!
sean_
·năm ngoái·discuss
Thank you for all your feedback, I made a new video which is more straight forward here: https://www.youtube.com/watch?v=fUTlF-KPEcw

Cloud Coding is a way to use claude code in a VM Sandbox.

The key feature of it is that you can script claude code to preform a sequence of prompts such as: - prompt 1: write a plan - prompt 2: implement the plan

This is extremely powerful and is what most "agents" apps are.

From using it everyday myself (mostly to build the site/coding agent), I've gotten better results than codex and at a much cheaper cost. I can more easily predict how much a run will cost because i control the workflow claude goes through.

The roadmap for this product isn't exactly clear:

I can compete head on with codex as a friendly UI general purpose coding agent or I could have it be made for workflows and integrations such as being triggering a template whenever a user commits to a repo with the template being a series of steps telling claude code to either create/run tests, lint, write docs, etc.

There's a huge range of possibilities so I'm looking for early users' feedback. This is why i've given each new user $5.00 of free credits to easily try it out!

Let me know what you guys think and I'm open to all feedback!
sean_
·năm ngoái·discuss
I added a youtube video explaining everything here https://youtu.be/_Tk2fnUF-Xc

While it's true the app requires access to your repos, many apps do and there's really no way around it unfortunately.
sean_
·năm ngoái·discuss
same idea, yes

this seems simpler and more straight forward though
sean_
·năm ngoái·discuss
But yes, the LLM could go wrong because it's simply a wrapper around scripting claude code in non-interactive mode.

the immediate use cases are more towards automatically creating tests and documentation, as well as other non-destructive actions

such as read-only mode: https://github.com/RVCA212/codesys/blob/main/examples/exampl...
sean_
·năm ngoái·discuss
The source code is 155 lines.

You can vibe through asking cursor if it has any bugs and let me know or create a PR!

But by having 155 lines of source code and through continuous use of the sdk, I haven't experienced any problems

Actually this comment just gave me the idea of creating a file of codesys doing just this in its own repo!
sean_
·năm ngoái·discuss
yea this boilerplate is super simple and effective. It's essentially mimicing how I use claude code and cursor, but with low level control.

A very cool thing I'm working on in this space is having an llm code with the codesys sdk, then run the code.

So imagine cursor coding a codesys file instead of doing the task directly so that it instead scripts claude code to do a sequence of actions and allows cursor/the user to simply analyze the results.

this also enables parralel claude code sessions which is super cool!
sean_
·năm ngoái·discuss
I never heard of Bob Loblaw, so no.

It's more a play on the phrase 'Code the thing that Codes the thing'
sean_
·năm ngoái·discuss
MCP Servers make it easy for LLMs to call tools. Since tools are mostly API calls, why not turn every API into an MCP Server tool?

This was my thesis behind building this feature into lmsystems.

The goal is simple:

- Add as many curl commands onto the site as possible and call them 'Tools'

- Combine them together into 'Servers' under 1 sse url

- Give every user the ability to do the same

I'm happy to say that we're nearly there !

I'm an indie hacker working on this completely alone so I would love feedback on how I can get the community more engaged and any features to add!

Also I'm interested in your thoughts on why this isn't going to work. What's going to kill this idea?