@dontbesilent2025

dbs-script-flow diagnoses short-form video scripts for audience drop-off risk by scanning paragraph transitions, information density, and delivery flow. identifies where viewers will swipe away and offers marked-up revisions.

dbs-ai-check detects ai fingerprints in chinese copy through 22 pattern signatures, outputting citation-based diagnostic reports. diagnosis is default mode, rewriting requires explicit user intent and iterative clarification of original intent.

dbs-chatroom orchestrates multi-expert dialogues on user topics by either accepting specified personas or recommending domain-matched experts based on perspective gaps. executes via parallel agent calls and synthesizes findings with moderator summary.

dbs-restore retrieves the most recent diagnostic snapshot for a project, presenting saved state (conclusions, rejected directions, pending assumptions) so users can resume work. integrates with dbs-save for persistent session storage across diagnostic workflows.

converts markdown to wechat-compatible html with 15 built-in style templates. includes interactive mode for style selection, preview generation, and output to browser-ready files with minimal external dependencies.

dbs-spread applies five classic communication theories (spiral of silence, uses and gratifications, framing, two-step flow, cognitive consistency) to decode why content resonates with audiences, mapping emotional substrates and discussion directions for chat rooms.

targeted-chatroom orchestrates multi-expert dialogue on user-specified topics, either accepting named participants or recommending domain specialists, then synthesizing viewpoints via parallel agent calls with structured prompts emphasizing authentic reasoning over formulaic responses.

dbs-report merges diagnostic snapshots from dbs-save into timestamped markdown reports. reads from ~/.dbs/sessions/, deduplicates conclusions, tracks problem evolution, and archives deliverables with version control.

dbs-skill-cleaner audits local agent skills for unauthorized commercial redirects, data exfiltration, and task hijacking using oecd ai principles. scans multiple vendor directories, reports findings by risk tier, then quarantines only after explicit user confirmation.

dbs-update syncs the official dbskill repository to supported agents without modifying user data or installing other skills. executes a single npm command after user confirmation.

chatroom-austrian orchestrates three-way debates between hayek, mises, and claude on austrian economics topics. the skill routes user questions to parallel agents, collects responses, and synthesizes judgments with blind spot coverage and actionable takeaways.

dbs-decision frames personal decision tracking as a four-layer knowledge system (facts, patterns, snapshots, open questions) with source labels and immutable records. five operational modes handle initialization, status updates, decision documentation, result backfill, and snapshot generation. no step-by-step procedures for core tasks.

dbs-hook diagnoses short-form video opening problems using a content-quality-first framework, then generates 10-15 hook variations across three methods. execution relies heavily on user compliance with quality gates, with limited procedural clarity on actual generation steps.

dbs-agent-migration standardizes multi-platform agent configuration (Claude Code, Codex, Grok) through structure audit and bridge generation, but lacks concrete procedure steps and failure handling.

dbs-content-system structures bulk local content into a reusable engineering system by auditing scale, building workflows, extracting units, and generating topic maps. targets users with substantial existing material.

dbs-knowledge is a folder-to-knowledge-base conversion tool that scans local file structures, generates navigation files (SOURCE_OF_TRUTH.md), handles file ingestion, retrieval and health checks. it avoids external databases and preserves original files unchanged.

dbs-slowisfast diagnoses which work phases merit deliberate slowness versus speed, grounded in friction-as-information and compounding assets. lacks executable step sequences and failure-mode thinking.

dbs-good-question is a framework for reformulating vague questions into structured problem statements with explicit objects, conflicts, and constraints. it provides extensive philosophy and phase definitions but lacks actionable step-by-step procedures and does not document failure modes or implementation details.

dbs-content is a content diagnostic framework for chinese-language creators, emphasizing writing quality and audience cognition over ai-generated prose. lacks actionable procedure.

dbs-action is a diagnostic conversational skill in chinese that applies adlerian psychology to identify psychological barriers to execution, not a cross-vendor database or action automation tool.

dbs-diagnosis presents philosophical axioms about business model diagnosis but lacks actionable procedure, decision logic, or concrete output contracts. not a skill, a manifesto.

dbs-xhs-title is a xiaohongshu title formula matcher claiming to select from 75 verified formulas indexed by psychological trigger categories, but provides neither the formula library nor actionable procedure.

dbs-save is a state serialization tool for diagnostic workflows, writing conversation conclusions and next steps to markdown snapshots. the implementation is sketched but lacks procedural specificity.

理论溯源与案例重释。审查 PPT、课程、文案或方法论中的经验判断，先判断命题是否成立，再匹配并核实可信理论，用该理论重释案例并标明边界。适用于给自创理论找依据、给 PPT 补理论来源、判断一句话能否上升为理论、检查是否硬套名人；聚焦单次理论锚定，不默认展开历史同构研究。

从抖音、小红书、微信视频号等平台链接或分享文案查询作品／账号数据并提取语音文字稿，按作者和标题归档为 Markdown。用户要求解析平台视频或取得数据、文字稿时使用；需配置 TikHub 与轻抖凭证，不负责本地成片章节导航或导航视频渲染。

通过需求分析把反复出现的问题制作成可安装、可分级验证的单个 Skill。用户要求做 Skill、把问题沉淀成 Skill、检查或测试候选 Skill 时使用；只有用户明确要求分享或发布时，才准备 GitHub 仓库并验证 `npx skills add`。

将单个 Skill 或 Skill 集合安装到通用 Agents、Claude Code、Codex、WorkBuddy、Grok、Hermes Agent、Kiro、Qwen Code、Cline 等 Agent。用户要求跨 Agent 安装、同步、查看、去重或卸载 Skill 时使用。

用 Jobs to Be Done 识别具体情境中用户想推进的进展、切换方案的力量和选择标准，解释产品、服务或内容为何被选择。用户要求分析顾客任务、购买动机、替代方案或明确用 JTBD 重写提示词时使用；普通任务验收标准澄清无需套用该框架。

面向创作者的内容发布风险检查。逐句检查自媒体、社交媒体和内容平台的标题、正文、图片文字、字幕、口播、评论、账号资料及视频画面，分别判断平台机器可能识别什么、内容本身存在什么实质问题，并给出具体位置和最小修改动作；需要明确视频语境边界时，额外生成真实准确的三行贴片小字。用户询问「检查敏感词」「发布前排雷」「这篇内容哪里可…

将单个 Skill 或 Skill 集合自动桥接到通用 Agents、Claude Code、Codex、WorkBuddy、Grok、Hermes Agent、Kiro、Qwen Code、Cline 等 Agent。用户要求跨 Agent 安装、同步、查看、去重或取消 Skill 链接时使用。

围绕复杂问题先挖掘相关行业、学科、作者与可信理论，再比较结构同构的历史成功、失败和反例，提炼带条件的标准答案。用户需要同时获得理论解释与历史比较，或明确询问历史上如何解决、有没有经典解法、标准答案时使用；只需给现有命题、案例、PPT 或方法论找理论依据时不使用。

诊断已经写成的稿件哪里没有打中目标受众，结合原句解释共鸣断点并提出具体删改建议。适用于稿件修订；不负责整体流量与商业价值评估，也不以解释已发布内容的传播机制为主要交付。

把课题拆成连续学习文章，并根据用户反馈调整后续深度、角度和节奏。用户要求系统学习一个主题、继续下一篇或按反馈调整课程时使用。

提取已有约束，只追问影响执行或验收的信息，把模糊愿望整理成可行动、可检查的交付目标。用户要明确做成什么、边界和验收标准时使用；不替代概念拆解、研究问题改写或顾客购买动机分析。

由哈耶克、米塞斯与 Claude 从奥派经济学视角展开多角色讨论。用户要求进入奥派聊天室或用奥派分析问题时使用。

用维特根斯坦与奥派经济学方法拆解模糊的商业概念。用户要求解释一个词、澄清概念边界或识别概念混用时使用。

用五重过滤法寻找值得模仿的对标，并排除主体差异造成的噪音。用户要求找对标、选择模仿对象或分析竞争参照时使用。

dontbesilent 商业工具箱主入口，提供新手教程、单任务编排、联网编号执行与隐藏款目录查询。用户第一次使用 dbskill、不知道该用哪个正式 Skill、输入方法视频的三位编号、查询所有隐藏款，或在任务完成后询问下一步时使用。

升级 dbskill 到最新版本

|