菜单
Establishes a performance baseline from existing ad account data. Use when starting work on a brand that has been running ads. Trigger on: 'audit the account', 'what's our baseline', 'how are we performing'. Requires dashboard or account access.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
"Imports raw transcripts (Fathom, Telegram, WhatsApp, email, free-form text — any format) into the Notion 'Real Customers' database. Creates one entry per external speaker identified, fills the 'Client input' phase properties (Name, Client relation, Transcript, dates, demographics if available), and leaves the 'AI extracted' fields for the Notion AI agent to populate. Use when the user pastes/uploads a call transcript, conversation log, sales call recording, or any customer interaction artifact they want ingested for downstream Persona work. Trigger on: 'import this transcript', 'ingest this call', 'add this prospect to Real Customers', 'push this Fathom into Notion'. Requires Notion MCP authenticated. Does NOT extract insights (sentiment, pain points, objections, summary) — that's the Notion AI agent's job, triggered separately."
"Builds a 3-layer persona from existing data and research. Use after brand-guidelines to deepen audience understanding before brief writing. Trigger on: 'who is our audience', 'build a persona', 'deep dive audience', 'who are we really talking to'. Produces a complete persona document ready to feed creative-brief and hook-writing."
"Builds a complete creative brief from research outputs. The brief defines the angle, awareness stage, hook direction, creative mechanic, format, specs, and success criteria. Use after audience-research and customer-reality. Trigger on: 'build a brief', 'creative brief', 'we're preparing a campaign', 'I want to run ads'. This is the central strategic document — nothing goes to production without it."
Runs a full Meta Ads account analysis in 7 steps using Windsor.ai as MCP data bridge. Pulls live campaign/creative/audience data directly into Claude and produces an actionable audit report: account health, campaign audit, creative lifecycle, angle & format performance, audience analysis, and prioritized action plan. Trigger on: 'analyze my Meta account', 'Meta Ads audit', 'which ads to tune', 'creative analysis Meta', 'Windsor analysis'. Requires Windsor.ai MCP connection.
Pilots the actual LAUNCH of ads on Meta via the official Meta/Facebook MCP, with hard guardrails at every step: full plan validated before any API call, everything created PAUSED, strict URL/UTM/creative rules, and a post-publication diff against the validated plan to catch Meta's silent auto-mutations. Use when assets are final and the campaign structure exists (see campaign-setup). Trigger on: 'launch these ads', 'push to Meta', 'publie les ads', 'mise en ligne campagne', 'put the campaign live'. Requires Meta MCP. Launching ads without this skill's checklist is how expensive Meta mistakes happen.
"Generates a validation agent .md file for any workflow step. The agent scores outputs 1-100 across 5 dimensions, provides actionable rewrites, and enforces hard violations. Use when adding quality gates to new or existing skills. Trigger on: 'create a validation agent', 'build a QA agent', 'I need a quality gate for [step]'. Output is a standalone agent-[name].md file ready for Project Knowledge upload."
| name | brand-health |
| description | Establishes a performance baseline from existing ad account data. Use when starting work on a brand that has been running ads. Trigger on: 'audit the account', 'what's our baseline', 'how are we performing'. Requires dashboard or account access. |
| metadata | {"version":"1.0.0","status":"beta","tags":["audit","performance","baseline","meta-ads"],"inputs":["Meta Ads dashboard access","minimum 30-day period"],"outputs":["baseline KPI report — top/bottom performers","strategic gaps"],"depends-on":["brand-guidelines"]} |
Confirm before starting:
brand-context.md available (otherwise launch brand-guidelines first)If dashboard access is not available: request a CSV export of campaign data.
This skill produces a baseline performance report before any new creative strategy is deployed.
Input: Meta Ads dashboard access + period (minimum 30 days, recommended 90 days) Output: Baseline report — KPIs, benchmarks, production capacity, constraints
Compare current KPIs against category benchmarks.
For benchmark reference: See references/brand-health.md
## Baseline Report — [Brand] — [Period]
### Key KPIs
| Metric | Current | Benchmark | Status |
### Top Performers
| Creative | Format | Primary KPI | Why It Works (hypothesis) |
### Bottom Performers
| Creative | Format | Primary KPI | Failure Hypothesis |
### Production Capacity
Creatives launched: [X] / [period]
Avg brief-to-launch: [X days]
### Strategic Gaps
[What's missing — formats, angles, awareness stages not covered]
01-audit/brand-alignment — run alongside for strategic alignment05-analysis/account-audit — deeper creative analysis after baseline set