# SPDX-License-Identifier: Apache-2.0 # Optional trusted manual run; the basic skill-checks.yml needs no model secret. # Use the release-generated copy with reviewed workflow and image pins. # The image uses the same default OSS policy for public and private repositories. # LLM content sharing is controlled by llm and the selected provider, not a profile. name: Skill checks with LLM on: workflow_dispatch: permissions: contents: read jobs: skills: uses: NVIDIA/SkillEvaluator/.github/workflows/evaluate-skills.yml@ba7779ad617c6444c98b6241a75bce92067e9d56 with: image: ghcr.io/nvidia/skillevaluator-tier1@sha256:e96e3aaad89bf78c4f3f8c621e88005d18bd7c485b29c0d7945864700e5944e6 paths: skills min-score: 70 security-baselines: false scan-mode: all llm: true llm-provider: openai secrets: LLM_API_KEY: ${{ secrets.LLM_API_KEY }}