Congrats on the launch, I'm a happy user of Dagger. One use case I've found for it besides CI was building an eval sandbox for testing different coding agents' performance using CLIs, APIs, and coding tasks.
It helps a lot with layering different permutations of dependencies (i.e. test Claude Code vs Codex vs Pi, or forward port access to the docs or no docs access) and you also benefit from the layer caching so evals take a lot less time.
One piece of constructive criticism I have for Dagger: while the API and CLI are beautifully made, the documentation and the webapp scare the normies and could be improved. I've also had some trouble with outdated documentation in the past but I'm sure 1.0 is meant to be a turning point in that with the new docs rework and all.
Thanks Pedro! Looks like we got buried on HN, oh well!
> One use case I've found for it besides CI was building an eval sandbox for testing different coding agents' performance using CLIs, APIs, and coding tasks.
Very nice! Would you be willing to do a video interview and show a demo for the community?
In my opinion evals and CI are converging into a common execution and data model for all pipelines involved in shipping your product. It's Continuous Engineering :)
> while the API and CLI are beautifully made, the documentation and the webapp scare the normies and could be improved.
Thanks for the feedback! Yes, we are working on the docs, it's not perfect yet but trending in the right direction I think. We are also betting that everyone will soon be asking their agent to setup Dagger, so we are investing heavily in making sure agents understand and appreciate Dagger, so they can set it up properly and explain it to their drivers. What a crazy world we live in!
Yes I'm actually gonna open-source it so I'll send you the repo, docs, and a recorded demo in a couple of weeks. We're already in contact on Discord too - I'm Pedro with the surfer pic.
I also see TypeSafe adopted Dagger which is awesome and hopefully inspires more confidence in the rest of the community. Best of luck!
Pedro_Ribeiro 1 day ago
Congrats on the launch, I'm a happy user of Dagger. One use case I've found for it besides CI was building an eval sandbox for testing different coding agents' performance using CLIs, APIs, and coding tasks.
It helps a lot with layering different permutations of dependencies (i.e. test Claude Code vs Codex vs Pi, or forward port access to the docs or no docs access) and you also benefit from the layer caching so evals take a lot less time.
One piece of constructive criticism I have for Dagger: while the API and CLI are beautifully made, the documentation and the webapp scare the normies and could be improved. I've also had some trouble with outdated documentation in the past but I'm sure 1.0 is meant to be a turning point in that with the new docs rework and all.
shykes 1 day ago
> One use case I've found for it besides CI was building an eval sandbox for testing different coding agents' performance using CLIs, APIs, and coding tasks.
Very nice! Would you be willing to do a video interview and show a demo for the community?
In my opinion evals and CI are converging into a common execution and data model for all pipelines involved in shipping your product. It's Continuous Engineering :)
> while the API and CLI are beautifully made, the documentation and the webapp scare the normies and could be improved.
Thanks for the feedback! Yes, we are working on the docs, it's not perfect yet but trending in the right direction I think. We are also betting that everyone will soon be asking their agent to setup Dagger, so we are investing heavily in making sure agents understand and appreciate Dagger, so they can set it up properly and explain it to their drivers. What a crazy world we live in!
Pedro_Ribeiro 1 day ago
I also see TypeSafe adopted Dagger which is awesome and hopefully inspires more confidence in the rest of the community. Best of luck!