Plan an A/B test by script: sample size per variant, days to run, stopping rules. "how many visitors per variant"
中文简介正在准备中。当前显示原文说明。
Mine raw signals (news, threads, sales notes) into content angles. "turn this industry news into ideas"
中文简介正在准备中。当前显示原文说明。
实测评分 未实测
查看安装方法安装前可以查看交给 AI 代理的指令内容。
The intelligence layer between "interesting" and "ours to say". Raw input comes from anywhere — a newsletter, a Reddit thread, three competitor posts, notes from yesterday's sales calls. The output is only the angles this brand has standing to make, each mapped to a pillar, with everything else explicitly dropped.
The mapping is the value. Any model can turn news into generic content ideas; the discipline is refusing the ideas that do not serve this brand's authority.
~/.claude-marketing/brands/{slug}/. No profile → stop: signal-mining
without pillars produces trend-chasing, which is the exact failure mode this
skill exists to prevent. Run /digital-marketing-pro:brand-setup first.# Signal mine — {brand}, {date}
## Angles ({n})
### A1. [The angle — the take, not the topic]
**From signal:** [one-line source reference]
**Pillar:** [brand pillar] **Timeliness:** [act this week / evergreen / expires ~date]
**Format:** [post / article / newsletter section / campaign hook]
**Why this brand:** [one sentence of standing — why this take is credibly ours]
## Dropped ({n})
- [signal] — [why: off-pillar / no standing / competitor's story to tell /
stale by the time we publish / compliance risk]
## Sourcing note
[Which signals carry claims that need verification before anything cites them —
route those through /digital-marketing-pro:verify-claims before drafting]
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Plan an A/B test by script: sample size per variant, days to run, stopping rules. "how many visitors per variant"
中文简介正在准备中。当前显示原文说明。
Write ad copy variants to each platform's specs, policy-checked and in brand voice. "write ad copy for Meta"
中文简介正在准备中。当前显示原文说明。
Add a custom MCP integration: scaffold the server, write .mcp.json, test it. "add a new MCP server"
中文简介正在准备中。当前显示原文说明。
Run a one-time AEO audit of six AI answer engines, scored per surface. "check our AI search visibility"
中文简介正在准备中。当前显示原文说明。
Plan AEO/GEO strategy to get cited by AI: content, entities, schema, 90-day plan. "how do we get cited by AI"
中文简介正在准备中。当前显示原文说明。
Build the agency dashboard across all client brands: health, pacing, risk. "how are all our clients doing"
中文简介正在准备中。当前显示原文说明。