It is an SDK for scripting Claude Code.
It's a lightweight (155 lines) and free wrapper around claude code
This is a big deal because it seems that using claude code and cursor has become largly repitive. My workflow typically goes like this:
Plan out my task into a file, then have claude code implement the plan into my code.
I'm actively building a product with this, but still wanted to make it OSS!
Use it now with `pip install codesys`
It's more a play on the phrase 'Code the thing that Codes the thing'
Probably "vibe-coded" in an hour and would likely have hundreds of bugs to be untrusted to use on my machine and doesn't remotely comply with 'clean code' architecture and test-driven development.
In the age of AI, it is even more important to have these tests for software that interacts with something that claims to reason and just adds unnecessary additional risk.
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!
the immediate use cases are more towards automatically creating tests and documentation, as well as other non-destructive actions
tarboreus•30m ago
sean_•10m ago
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!