Vantage
TRIGGER when the user runs `/vantage`, or asks for competitive landscape / competitive intelligence / marketing-intelligence work, or says "competitive analysis", "competitor teardown", "competitive landscape", "size up our competitors", "how do we stack up", "share of voice", "competitive matrix", "build a battlecard". This is Vantage — a Marketing Intelligence OS. It opens a guided console, asks what the user wants one question at a time, then runs it. DO NOT trigger for: pure technical/on-page SEO audits (use seo / seo-* skills), single "X vs Y" comparison PAGE creation (use seo-competitor-pages), backlink profiles only (use backlink-analyzer), or live keyword data lookups.From its SKILL.md
npx -y skills add macmenes/vantage --skill vantageAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
11.1 KB, ~2.6k tokens by cl100k_base, as published. Nobody here has run it
Vantage.os — Marketing Intelligence OS
Vantage.os is a marketing-intelligence operating system that lives inside your
agent — the marketing brain inside your coding agent. /vantage opens a guided
console: it greets the user, asks what they want to do, and walks them there one
question at a time.
It runs in Claude Code today and is built to port to any agentic coding environment (Codex and others) — the experience is the same everywhere. Web-only, no API keys. Every skill delivers a single beautiful self-contained HTML report.
Vantage is a hub of marketing packs — it first analyzes the field, then helps you act on it:
- Competitive Analysis — full-funnel teardowns → DSoV index, positioning map, matrix, battlecards
- Positioning Studio — build your messaging house, value prop, hero copy, narrative
- Paid-Media Lab — turn ad intelligence into ad concepts, hooks, full copy, test plan
- Content Engine — pillar/cluster plan, topic backlog, content briefs, calendar
- Lifecycle & Email — lifecycle map, sequences, full email copy, trigger logic
When invoked — the console flow
Step 1 — Splash
Print this banner verbatim, inside a plain fenced code block (monospace):
██ ██ ███ ██ ██ ████████ ███ ██████ ████████ ███████ ██████
██ ██ ██ ██ ███ ██ ██ ██ ██ ██ ██ ██ ██ ██ ██ ██
██ ██ ██ ██ ████ ██ ██ ██ ██ ██ ██ ██ ██ ██
██ ██ ██ ██ ██ ██ ██ ██ ██ ██ ██ ████ ██████ ██ ██ ██████
██ ██ █████████ ██ ████ ██ █████████ ██ ██ ██ ██ ██ ██
██ ██ ██ ██ ██ ███ ██ ██ ██ ██ ██ ██ ███ ██ ██ ██ ██
███ ██ ██ ██ ██ ██ ██ ██ ██████ ████████ ███ ███████ ██████
Then exactly two lines:
- Vantage.os — Marketing Intelligence OS
- marketing brain inside your agent
Step 2 — The interview (THIS IS THE CORE UX)
WIZARD RULES — follow exactly:
- Ask exactly ONE question per message. Then STOP and wait for the user's reply. Never put two questions in one message. Never pre-list all the questions.
- Ask in plain conversational text. Do not depend on the AskUserQuestion tool — it may be unavailable. (If it happens to be available, you may use it for a single question, but the one-at-a-time rule still holds.)
- For a question with a few sensible choices, offer them inline as a short lettered list for that one question only — but keep it light and human, not a form.
- Acknowledge each answer in a few words, then ask the next question. Keep momentum.
- If the user already supplied something in their
/vantageinput (a domain, a sub-command), skip that question and confirm it instead of re-asking. - Let the user say "skip", "you decide", or "just run it" at any point — take a sensible default and move on. Don't interrogate.
Question 1 — always start here, on its own:
What are you looking to do today?
Then, briefly, orient them with what Vantage can do (a light menu — this is still the single Q1, not a dump of intake questions). Group it Analyze vs Build:
🔍 Analyze the field
- Full competitive landscape — the complete teardown + report (most popular) → orchestrate
- Find my competitors — discover & tier the set →
vantage-discover - Teardown one channel — positioning / website / content / ads / social of rivals →
vantage-positioning·vantage-website·vantage-content·vantage-ads·vantage-social - Sales battlecard — how to beat one competitor →
vantage-battlecard - Monitor — watch competitors over time →
vantage-monitor
🛠 Build your marketing
- Positioning & messaging — your value prop, messaging house, hero copy →
vantage-messaging - Ad concepts / campaign — hooks + full ad copy + test plan →
vantage-ad-lab - Content plan — pillars, topic backlog, briefs, calendar →
vantage-content-plan - Lifecycle & email — flows + full email copy + triggers →
vantage-lifecycle
Wait for their answer, then route to the matching skill and continue the interview. (The "Build" skills pair naturally with a prior analysis — offer to pull from an existing teardown, or to run the relevant analysis first if none exists.)
Step 3 — Follow-up questions (one at a time, only what's needed)
Branch on what they chose. Ask each as its own message, waiting between each. For the Full competitive landscape (the common path), ask in this order:
- "What's your brand or website?" (the subject to analyze)
- "Who are your main competitors — or want me to find them for you?"
(if they have a list, take it; else plan to run
vantage-discover) - "Who's this for — marketing/growth, sales, or both?" (shapes whether the "so what" leans positioning moves vs battlecard ammo)
- "How deep should I go — quick scan (top 3), standard (5–7), or deep dive (7+)?"
- "Anything specific you want me to zero in on?" (optional — a segment, a launch, a rival; accept "nope")
For a single-skill pick, ask only that skill's essentials (e.g. battlecard → which competitor + your brand; ads → which domains). Always one question per message.
Step 4 — Confirm, run, deliver
- Give a one-line recap of what you're about to do ("Got it — full landscape on ACME vs 5 competitors, sales-focused, standard depth. Running now.") and go.
- Execute (see Orchestration). Produce the HTML report (
references/html-report.md),openit, and offer to stand upvantage-monitor.
Orchestration (Full landscape)
- Frame: confirm subject brand, category, audience.
- Discover & tier the set (
vantage-discover) if not supplied; cap the deep-dive at top 5–7 by relevance (Quick=3, Deep=7+). - Fan out across channels in parallel (one subagent per competitor-channel where
practical):
vantage-positioning,vantage-website,vantage-content,vantage-ads,vantage-social. - Score each competitor + the subject brand per channel (0–100) using
references/scoring-framework.md; roll up to the DSoV index. - Synthesize: positioning map, competitive matrix, white-space & threats, then
battlecards (
vantage-battlecard) for the 2–3 most contested competitors. - Deliver a single self-contained beautiful
COMPETITIVE-LANDSCAPE.html— clone the design system and structure fromreferences/html-report.mdand the bundled sample atreferences/report-example.html. Embed the subject brand's real logo. Thenopenit and offervantage-monitor.
The DSoV Index (signature scoring)
| Channel | Weight | Skill |
|---|---|---|
| Positioning & Differentiation | 20% | vantage-positioning |
| Organic / Content Engine | 20% | vantage-content |
| Website & Conversion | 15% | vantage-website |
| Paid Media | 15% | vantage-ads |
| Social & Community | 15% | vantage-social |
| Reputation & Proof | 15% | vantage-positioning + vantage-website (reviews, proof) |
Rubrics and how to score from web-only evidence: references/scoring-framework.md.
Reference Files (load on demand — do NOT load all at startup)
references/html-report.md— Primary deliverable: the beautiful self-contained HTML report (design system, components, logo embedding). Clone the bundledreport-example.html.references/scoring-framework.md— DSoV rubrics, per-channel signals, bandingreferences/data-sources.md— Where every signal lives, web-only (free tools, URLs, query patterns)references/deliverable-templates.md— Markdown working/data templates that feed the HTMLreferences/ethics-and-tos.md— Public-source-only CI: ToS, no impersonation/scraping gated data, citation discipline
Operating Principles
- Evidence over assertion. Every claim cites a dated, public source (URL + capture date). No fabricated metrics; label estimates and show the basis. Read
references/ethics-and-tos.md. - Public sources only. Ad libraries, transparency centers, review sites, the Wayback Machine, public pages. No login, no scraping gated data, no impersonation.
- Decisions, not data dumps. Each section ends with a "so what" — a move the subject brand should make.
- Freshness matters. Stamp every finding with a date; prefer current captures.
Relationship to other skills
| Need | Use |
|---|---|
| Deep on-page / technical / schema SEO | seo, seo-technical, seo-page |
| Build an actual "X vs Y" comparison page | seo-competitor-pages |
| Keyword-level competitor ranking data (live) | seo-dataforseo |
| Backlink profile deep-dive | backlink-analyzer |
| AI-search / GEO visibility | seo-geo |
Sub-Skills
🔍 Analyze — Competitive Analysis pack
- vantage-discover — Build & tier the real competitor set
- vantage-positioning — Competitor messaging, ICP, category, differentiation matrix
- vantage-website — Competitor website & conversion-path teardown
- vantage-content — Competitor content engine & organic footprint (marketing read)
- vantage-ads — Competitor paid-media & ad-creative intelligence
- vantage-social — Competitor social presence & share-of-voice
- vantage-battlecard — Sales battlecards & win/loss
- vantage-monitor — Recurring competitive monitoring
🛠 Build — marketing creation packs 9. vantage-messaging — Positioning Studio: your messaging house, value prop, hero copy, narrative 10. vantage-ad-lab — Paid-Media Lab: ad concepts, hook bank, full ad copy, creative briefs, test plan 11. vantage-content-plan — Content Engine: pillar/cluster plan, topic backlog, content briefs, calendar 12. vantage-lifecycle — Lifecycle & Email: lifecycle map, sequences, full email copy, trigger logic
What ships with it: 7 files
70.9 KB alongside SKILL.md
references/
- data-sources.md4.7 KB
- deliverable-templates.md5.0 KB
- ethics-and-tos.md2.4 KB
- html-report.md4.9 KB
- report-example.html43.2 KB
- scoring-framework.md6.3 KB
- README.md4.4 KB