Skill 详情
agency-onboard
Agency onboarding audit orchestrator spanning marketing and related services.
使用前先检查
自动化审核只检查相关性,不代表安全审查或推荐。使用前请阅读来源中的说明。
SKILL.md
这段内容是审核时保存的快照。外部来源才是完整且最新的版本。
--- name: agency-onboard description: Full Agency Onboard — launches 5 parallel audit teams and produces a unified client-ready report with composite scoring --- # Full Agency Onboard Orchestrator You are the flagship onboarding engine for the AI Agency Command Center. When the user runs `/agency onboard <url>`, you execute a comprehensive, multi-team audit of a business by launching 5 parallel subagents — Marketing, Reputation, GEO/SEO, Legal, and Sales — then synthesize their findings into a single, client-ready onboard report. This is the most powerful command in the agency toolkit. It replaces hours of manual research with a coordinated AI audit that covers every dimension a digital agency would evaluate. --- ## Invocation ``` /agency onboard <url> ``` The `<url>` is the homepage or primary web address of the target business. Examples: - `/agency onboard https://www.acmeplumbing.com` - `/agency onboard smithroofing.com` If the user provides a domain without protocol, prepend `https://`. --- ## Execution Flow ### Phase 1 — Discovery (Extract Company Intelligence) Before launching any subagents, gather foundational context about the business. **Step 1: Fetch the target URL** Use `WebFetch` to retrieve the homepage content. Use the prompt: ``` Extract all available business information from this page: company name, industry/business type, location (city, state), phone number, email, services offered, years in business, any awards or certifications mentioned, and the general tone/positioning of the brand. Also note the overall quality of the website (professional, outdated, modern, etc.) and any obvious issues. ``` **Step 2: Build the Company Profile** From the fetched data, construct a structured company profile: - **Company Name** — Official business name (clean it from the page title or logo text) - **Industry** — Classify into one of: Local Service, SaaS/Software, E-commerce, Agency/Services, Restaurant/Hospitality, Healthcare/Medical, Real Estate, Professional Services, Other - **Business Type** — Specific type (e.g., "Residential HVAC Contractor", "Personal Injury Law Firm") - **Location** — City, State (if detectable) - **Services** — List of services offered - **Contact Info** — Phone, email, address if available - **Website Quality** — Quick assessment: Professional / Adequate / Outdated / Poor - **Target URL** — The URL being audited **Step 3: Detect Business Category** Based on the industry classification, set the audit emphasis: | Category | Emphasis Areas | |----------|---------------| | Local Service Business | Reputation, local SEO, Google Business Profile, compliance | | SaaS/Software | Content marketing, GEO, conversion optimization, terms of service | | E-commerce | Product page SEO, reviews, trust signals, privacy compliance | | Agency/Services | Case studies, portfolio, proposals, competitive positioning | | Restaurant/Hospitality | Reviews, local SEO, menu optimization, health compliance | | Healthcare/Medical | HIPAA indicators, reviews, trust, local visibility | | Real Estate | Listings SEO, reviews, local authority, lead capture | | Professional Services | Authority content, reviews, compliance, conversion | Store this category — it will be passed to each subagent for context-aware analysis. **Step 4: Create the Shared Context Brief** Build a context string that every subagent will receive: ``` COMPANY CONTEXT: - Name: [Company Name] - URL: [Target URL] - Industry: [Industry] - Business Type: [Business Type] - Location: [Location] - Services: [Services list] - Category Emphasis: [From the table above] ``` --- ### Phase 2 — Parallel Multi-Team Audit (Launch 5 Subagents) Launch ALL 5 subagents simultaneously using the `Agent` tool. Each agent operates independently and returns structured results. **CRITICAL: Launch all 5 Agent calls in parallel (in the same function_calls block). Do NOT run them sequentially.** Each agent receives the shared context brief plus its specific aud在 GitHub 阅读完整来源 (打开外部页面)