Test, fold or refute hypotheses beside the knowledge record, and reconcile records across branches. Use when open or check say hypotheses wait or an id is CONTESTED, when a write was refused into a hypothesis, before merging a branch whose record changed, and when a pull request's dry run or remeasure is red. Covers consolidate --dry-run, the fold, --refute, --from, same and distinct at the fold, and remeasure.
这个条目提供什么？
consolidate 是 ChatHome Skills 收录的 技能 条目，来源为 GitHub Skill Search，质量分 58/100。 # Consolidate A hypothesis is a claim the base does not yet hold: a refused reading, a proposal not yet approved, a branch's record, a what-if. Nothing here dec. 作为 agent 技能，它把一组可复用的任务指令打包成 Codex 和 Claude Code 等 AI 编程 agent 可直接加载的格式，无需为该任务手写胶水代码。 该条目由 ChatHome Skills 的每日抓取从官方 registry 和 GitHub 发现查询中收录，每次运行都会重新核验仓库事实、新鲜度信号和安装元数据。 质量分综合了官方 registry 收录情况、可用安装命令、许可证完整性、仓库活跃度和 GitHub star 趋势，完整算法说明见来源页。安装命令已列于上方，可直接复制使用。 在将 consolidate 加入生产 agent 工作流前，请核对上方的最近更新日期和仓库状态——下方事实快照是最近一次抓取的时点值。
某个来源结构上无法提供的维度（registry 不提供 star 和许可证）不计入该条目的平均分，而不是记为 0 分，因此不同来源的条目仍可横向比较。
安全信号
扫描范围: 安装元数据与已收录的文档正文。
基于安装元数据和有界技能正文的自动风险信号，不是完整代码安全审计，也不会递归审查源码或依赖。
发现备注
# Consolidate A hypothesis is a claim the base does not yet hold: a refused reading, a proposal not yet approved, a branch's record, a what-if. Nothing here decides for a person; t
Repository-grounded threat modeling that enumerates trust boundaries, assets, attacker capabilities, abuse paths, and mitigations, and writes a concise Markdown threat model. Trigger only when the user explicitly asks to threat model a codebase or path, enumerate threats/abuse paths, or perform AppSec threat modeling. Do not trigger for general architecture summaries, code review, or non-security design work.