Multi-agent mathematical problem-solving & verification frameworks for DeepSeek Harness — FOUR agent presets in one install: vibe-math-v2 (probability-driven, classic: qs.json + Propos knowledge base + explorer→solver→review/debate verdict), vibe-math-v3

Governance standard and reference toolset for LLM-maintained knowledge corpora

Governance standard and reference toolset for LLM-maintained knowledge corpora