Contribution policy — third-party hosted inference endpoints in nim-skills

[boldakov]

Hi BioNeMo team,

We are building Genomic Intelligence, a genome-scale model layer for regulatory genomics workflows. It is served via REST API and MCP, spanning a range of regulatory DNA analysis tasks - promoter and enhancer activity prediction, chromatin accessibility, splice-site prediction, gene expression prediction, and sequence annotation.

The nim-skills/ pattern (a SKILL.md plus a bearer-auth REST call) is a clean fit, so we'd like to contribute a genomic-intelligence-nim skill following your existing conventions. Before we invest, one scoping question:

The existing nim-skills/ target NVIDIA-operated endpoints. Is a third-party hosted inference endpoint run by us in scope for this toolkit?

If yes, we'll open a small PR with a single task first (promoter activity prediction) and expand from there toward a fuller regulatory-variant workflow. If hosted third-party endpoints aren't in scope, we'd welcome your guidance on the integration path you'd prefer.

We're happy to match your preferred skill format and licensing. Would this kind of contribution be welcome?

[boldakov]

Hi BioNeMo team — just following up on this.

We are happy to keep the first contribution very small and aligned with the existing nim-skills/ structure: one genomic-intelligence-nim skill with a single promoter activity prediction task, references, and evals.

The main scoping question is whether third-party hosted inference endpoints are acceptable in this repo. If not, we would appreciate guidance on the preferred path — for example, whether this should live as an external/community skill, a separate integration, or another format.

Thanks!