Hi, I am Haseeb. I am building GuardStep, an experimental language and runtime for typed AI workflows.
The idea is to keep tool permissions, output validation and execution limits explicit in the workflow, instead of spreading that logic across an application.
There's a working npm alpha and a document-Q&A example. The implementation is in TypeScript. It's still early and not production-ready.
A few developers have already contributed fixes. If you'd like to get involved, there are small documentation tasks as well as compiler and runtime work.
Two places to start:
Update the README example to match the current syntax (#49):
https://github.com/haseebahmed248/GuardStep/issues/49Make the npm package easier to discover and get started with (#30):
https://github.com/haseebahmed248/GuardStep/issues/30
Please check whether an issue is already claimed and leave a comment before starting. Trying the example and reporting where you get stuck would also help.
Repository: https://github.com/haseebahmed248/GuardStep
npm: https://www.npmjs.com/package/guardstep
Top comments (0)