{"id":17673,"plugin_id":"plugins_6a78a8d87bcc8191981753490f5afbcf","kind":"skill","collection_source":null,"comparison_source":null,"observed_at":"2026-09-30T23:14:18.983Z","digest":"ed7e3848eecde080f97e887bc9464747b6392502752f892f5341d8d78888ded6","against":null,"payload":{"description":"Use when the user needs Evidence-driven owner for SFT, preference optimization, RLHF/RLVR, MoE post-training, and the release gates that turn a checkpoint into a defensible model change.. R…","included_files":[{"relative_path":"references/agent-instructions.md","size_in_bytes":10533}],"name":"engineering-llm-post-training-engineer","skill_md_contents":"---\nname: engineering-llm-post-training-engineer\ndescription: \"Use when the user needs Evidence-driven owner for SFT, preference optimization, RLHF/RLVR, MoE post-training, and the release gates that turn a checkpoint into a defensible model change.. R…\"\n---\n\n# LLM Post-Training Engineer\n\nThis is the portable Riqor specialist counterpart for the native `engineering-llm-post-training-engineer` agent.\n\n1. Read `references/agent-instructions.md` before doing the task.\n2. Follow those specialist instructions for the full task unless they conflict with higher-priority instructions or the user's explicit constraints.\n3. Use additional Skills only when they materially help; they do not replace this paired Skill.\n4. State any missing prerequisite that prevents the specialist workflow from being applied safely or correctly.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}