Update skill docs/evals to short Docker tag scheme (latest-cu12 / latest-cu13)
@ramakrishnap-nv is already working on this.
Since Jul 21, 2026.
- Dominant language
- Cuda
- Stars
- 1k
- Forks
- 233
- Avg merge
- 4d 4h
- Merged PRs (30d)
- 95
Description
Context
We are simplifying the published Docker image tags to add short cu<major> aliases (latest-cu12, latest-cu13, <version>-cu12/cu13) alongside the existing fully-qualified …-cuda<minor>-py<py> tags. The CI manifest publish (ci/docker/create_multiarch_manifest.sh), README.md, the docs install-selector, the NIM operator examples, and the gRPC docs have been updated to lead with the short tags.
The agent skills still reference the old nvidia/cuopt:latest-cuda12.9-py3.13 tag and should be updated to stay consistent (these go through the skill-evolution / CI validation flow, so tracking separately).
Files to update
Skill doc examples:
skills/cuopt-install/SKILL.md(docker pull/run examples)skills/cuopt-server-api-python/SKILL.md(docker run example)skills/cuopt-install/references/verification_examples.md(docker run example)
Skill eval fixtures (ground-truth asserts the recommended tag):
skills/cuopt-install/evals/evals.jsonskills/cuopt-install/benchmark/evals.json
Suggested change
Replace latest-cuda12.9-py3.13 (and similar) with latest-cu12 / latest-cu13. Update eval ground-truth and acceptance criteria so an agent following the new convention passes.
Note: short aliases only exist for releases built after the manifest change lands (26.6.0+).
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.