Stats
Actions
Tags
Manages recursive decision rollouts with consistency markers and gating. Defaults to safe modes unless explicitly approved. Supports high-dimensional search and ensemble comparisons.
How this skill is triggered — by the user, by Claude, or both
Slash command
/everything-claude-code:recursive-decision-ledgerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
当用户试图通过重复推出或"Prime Gauss"风格的递归提示来强制更深入的计算时使用此技能。保留有用的部分：重复试验、先验记忆、新鲜信息和显式标记。移除不安全的的部分：假装循环证明确定性。
当用户试图通过重复推出或"Prime Gauss"风格的递归提示来强制更深入的计算时使用此技能。保留有用的部分：重复试验、先验记忆、新鲜信息和显式标记。移除不安全的的部分：假装循环证明确定性。
每次推出应记录：
首选 JSONL 用于仅追加账本，Markdown 用于人类摘要。
包含一个紧凑的一致性标记：
集成匹配先前获胜者：true
递归匹配先前获胜者：false
最新推出匹配：true
允许实时提升：false
原因：重放和新鲜度门控未满足
对于交易、资本分配、生产部署、迁移或破坏性操作，递归置信度不是批准。
默认为模拟、空运行、只读、预览或分阶段模式，除非用户明确批准实时操作且仓库/服务门控支持。
仅在以下情况下提升：
以决策开头，而非戏剧：
推出 15 完成。先前获胜者仍然成立，但优势劣化了 17%。
状态：观察，非实时。下一个门控：20 次重放填充且新鲜订单簿年龄低于阈值。
npx claudepluginhub aaione/everything-claude-code-zhManages recursive decision processes with explicit rollouts, coherence marks, and promotion gates for repeated trials or local-optima exploration.
Evaluates competing approaches through independent assessment, reasoning-out-loud advocacy, and confidence-calibrated thresholds to produce coherent decisions from multiple reasoning paths.
Auto-Decision Engine: iterative decision simulation using autoresearch principles and a persona council. Decomposes decisions, generates competing hypotheses, simulates first/second-order effects with probabilities, critiques via anonymized peer review, and iterates until insights stabilize mechanically. The output is a possibility map — hypotheses, effects, council disagreements, adversarial scenarios, assumptions — with a recommendation synthesized at the end, not as the product.