To access TBs of (mostly neural) data available online. Although base claude has good ideas on how to use them, skill might instruct on basic searching through https://registry.datalad.org and explicitly list some major available resources such as https://github.com/OpenNeuroDatasets/, https://github.com/dandisets and so on.
Just an idea for now
edit: but then major advancement could be to facilitate compute provenance tracking etc via datalad run and containers-run, see more on that in https://handbook.datalad.org/en/latest/basics/101-108-run.html or (although ai generated and not yet curated yet) https://myyoda.github.io/principles-examples/examples/datalad-run-provenance/ for basic demos or recentish blog post by @just-meng https://blog.datalad.org/posts/git-worktree-workflow/
To access TBs of (mostly neural) data available online. Although base claude has good ideas on how to use them, skill might instruct on basic searching through https://registry.datalad.org and explicitly list some major available resources such as https://github.com/OpenNeuroDatasets/, https://github.com/dandisets and so on.
Just an idea for now
edit: but then major advancement could be to facilitate compute provenance tracking etc via
datalad runandcontainers-run, see more on that in https://handbook.datalad.org/en/latest/basics/101-108-run.html or (although ai generated and not yet curated yet) https://myyoda.github.io/principles-examples/examples/datalad-run-provenance/ for basic demos or recentish blog post by @just-meng https://blog.datalad.org/posts/git-worktree-workflow/