[Agent] Add NPU main2main skill

[gcanlin]

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.

Purpose

This skill guides the process of upgrading vllm-omni's NPU model runners to align with the latest vllm-ascend codebase while preserving omni-specific enhancements. The NPU runners are designed to run omni multimodal models (like Qwen3-Omni, Bagel, MiMoAudio) on Ascend NPUs.

Test Plan

/vllm-omni-npu-upgrade

Test Result

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the test style doc
• The test results. Please paste the results comparison before and after, or the e2e results.
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. Please run mkdocs serve to sync the documentation editions to ./docs.
• (Optional) Release notes update. If your change is user-facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)

[chatgpt-codex-connector]

Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits.
Credits must be used to enable repository wide code reviews.

[gcanlin]

@hsliuustc0106 PTAL

[hsliuustc0106]

BLOCKER scan:

• Correctness: PASS (skill content looks comprehensive and well-structured)
• Reliability/Safety: PASS
• Breaking Changes: PASS (adds new skill, no changes to core code)
• Test Coverage: ISSUES (Test Results section is empty)
• Documentation: ISSUES (Test Plan is just a path, not actual commands)
• Security: PASS

OVERALL: 2 BLOCKERS FOUND

VERDICT: REQUEST_CHANGES

Blocker 1: SKILL.md exceeds 300-line limit

The SKILL.md file is 300 lines, which exceeds the 300-line limit for skill bodies (the validation script enforces < 300 lines). Please condense the skill to under 300 lines by:

• Moving detailed tables to reference files
• Reducing verbose examples
• Keeping only the essential workflow steps in SKILL.md

Blocker 2: Incomplete PR description

The Test Plan section just has a path (/vllm-omni-npu-upgrade) without actual test commands or verification steps. The Test Results section is completely empty.

Please provide:

1. Actual test commands to verify the skill works
2. Test results showing the skill was successfully tested

The skill content itself is comprehensive and well-structured - good documentation of the NPU upgrade workflow, omni-specific blocks, and translation patterns. Just need to address the length limit and test documentation.

[hsliuustc0106]

any test results for this skill?

[gcanlin]

any test results for this skill?

Let me create a PR to try it.