discover-brand by anthropics/knowledge-work-plugins
npx skills add https://github.com/anthropics/knowledge-work-plugins --skill discover-brand协调跨企业平台的品牌材料自主发现。此技能协调 discover-brand 代理搜索已连接平台(Notion、Confluence、Google Drive、Box、Microsoft 365、Figma、Gong、Granola、Slack),对来源进行分类,并生成一份包含开放性问题的结构化发现报告。
开始之前,简要解释即将发生的事情,让用户了解预期情况:
"品牌发现的工作方式如下:
.claude/brand-voice-guidelines.md 中。在该步骤之前,不会写入任何内容。搜索通常需要几分钟,具体取决于连接了多少个平台。准备好开始了吗?"
等待用户确认后再继续。如果他们对流程有疑问,请先回答。
如果存在 .claude/brand-voice.local.md 文件,则读取它。提取:
如果不存在设置文件,则使用所有已连接的平台和标准搜索深度进行。
在确认范围之前,检查实际连接了哪些平台并进行分类:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
文档平台(品牌指南、风格指南、模板和演示文稿所在之处):
补充平台(对模式有价值,但不是品牌文档存储之处):
应用以下规则:
如果零个文档平台已连接:停止。 告诉用户:"您没有连接任何文档存储平台(Google Drive、SharePoint、Notion、Confluence 或 Box)。品牌指南和风格指南几乎总是存在于其中之一。请在运行发现之前至少连接一个平台。Gong/Granola/Slack 文字记录是有价值的补充,但不太可能包含正式的品牌文档。"
如果没有连接 Google Drive 且没有连接 Microsoft 365 且没有连接 Box:警告(但继续):"您的主要文件存储平台(Google Drive、SharePoint、Box)均未连接。品牌文档经常存在于这些平台上。发现将继续使用[已连接平台],但结果可能存在重大空白。请考虑连接 Google Drive 或 SharePoint。"
如果总共只连接了一个平台:警告(但继续):"仅连接了[平台]。发现功能在连接 2 个以上平台以进行跨来源验证时效果最佳。来自单个平台的结果置信度得分较低。"
在启动发现之前,确认:
保持简洁——一个问题,而不是一份问卷。
通过 Task 工具启动 discover-brand 代理。提供:
代理自主执行 4 阶段发现算法:
当代理返回时,向用户呈现报告并附上摘要:
呈现报告后,提供:
/brand-voice:generate-guidelines当发现代理遇到无法解决的模糊情况时,会出现开放性问题:
每个开放性问题都包含代理建议。将问题呈现为"确认或覆盖"——而不是死胡同。
有关详细的发现模式和算法,请查阅:
references/search-strategies.md — 特定于平台的搜索查询、按平台的查询模式以及最大化发现覆盖范围的技巧references/source-ranking.md — 来源类别定义、排序算法权重和分类决策标准每周安装量
74
代码仓库
GitHub 星标数
8.8K
首次出现
14 天前
安全审计
安装于
codex71
opencode69
github-copilot69
kimi-cli69
amp69
gemini-cli69
Orchestrate autonomous discovery of brand materials across enterprise platforms. This skill coordinates the discover-brand agent to search connected platforms (Notion, Confluence, Google Drive, Box, Microsoft 365, Figma, Gong, Granola, Slack), triage sources, and produce a structured discovery report with open questions.
Before starting, briefly explain what's about to happen so the user knows what to expect:
"Here's how brand discovery works:
.claude/brand-voice-guidelines.md in your working folder once you approve them. Nothing is written until that step.The search usually takes a few minutes depending on how many platforms are connected. Ready to get started?"
Wait for the user to confirm before proceeding. If they have questions about the process, answer them first.
Read .claude/brand-voice.local.md if it exists. Extract:
If no settings file exists, proceed with all connected platforms and standard search depth.
Before confirming scope, check which platforms are actually connected and classify them:
Document platforms (where brand guidelines, style guides, templates, and decks live):
Supplementary platforms (valuable for patterns, but not where brand docs are stored):
Apply these rules:
If zero document platforms are connected : Stop. Tell the user: "You don't have any document storage platforms connected (Google Drive, SharePoint, Notion, Confluence, or Box). Brand guidelines and style guides almost always live on one of these. Please connect at least one before running discovery. Gong/Granola/Slack transcripts are valuable supplements but unlikely to contain formal brand documents."
If no Google Drive AND no Microsoft 365 AND no Box : Warn (but proceed): "None of your primary file storage platforms (Google Drive, SharePoint, Box) are connected. Brand documents frequently live on these platforms. Discovery will proceed with [connected platforms], but results may have significant gaps. Consider connecting Google Drive or SharePoint."
If only one platform total is connected : Warn (but proceed): "Only [platform] is connected. Discovery works best with 2+ platforms for cross-source validation. Results from a single platform will have lower confidence scores."
Before launching discovery, confirm:
Keep this brief — one question, not a questionnaire.
Launch the discover-brand agent via the Task tool. Provide:
The agent executes the 4-phase discovery algorithm autonomously:
When the agent returns, present the report to the user with a summary:
After presenting the report, offer:
/brand-voice:generate-guidelines using discovery report as inputOpen questions arise when the discovery agent encounters ambiguity it cannot resolve:
Every open question includes an agent recommendation. Present questions as "confirm or override" — not dead ends.
For detailed discovery patterns and algorithms, consult:
references/search-strategies.md — Platform-specific search queries, query patterns by platform, and tips for maximizing discovery coveragereferences/source-ranking.md — Source category definitions, ranking algorithm weights, and triage decision criteriaWeekly Installs
74
Repository
GitHub Stars
8.8K
First Seen
14 days ago
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
codex71
opencode69
github-copilot69
kimi-cli69
amp69
gemini-cli69
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
54,900 周安装
品牌声音准则生成工具 - 从文档、对话和报告自动创建LLM就绪的品牌指南
311 周安装
Axiom 仪表板构建指南:设计决策优先的监控仪表板与数据可视化
311 周安装
Google Ads Manager 技能:广告系列管理、关键词研究、出价优化与效果分析
311 周安装
Telegram机器人开发教程:构建AI助手、通知系统与群组自动化工具
311 周安装
AI图像生成提示词优化指南:DALL-E、Midjourney、Stable Diffusion提示工程技巧
311 周安装
AI协作头脑风暴工具 - 将想法转化为完整设计规范,支持代码模板与项目管理
311 周安装