For Agents · MCP

Your task.
The right skill.

The tunnel uses the same cards as the catalogue. Browse only as deep as needed — or load a broad bundle immediately.

No local installationNo M11 loginRead-only

Broad task

Load 5–10 and go

SEO, Sales, Agents or another broad area → one bundle call → work.

MCP endpoint: https://skills.m11.ch/mcp

Read-only access to published skills. Default 8, maximum 10 skills / 120,000 characters.

Task Packs

Marketing Foundation

Compact two-skill starter: clarify positioning and choose a lead magnet. Use Marketing Launch for the broader eight-skill go-to-market workflow.

Marketing Launch

Build an evidence-led marketing plan from ICP and competition through positioning, campaigns, growth and measurement.

Agent Audit Essentials

Diagnose architecture and context, plan agent-team responsibilities, then organize project context and session handoffs. Memory and cost-runtime reviews remain outside this pack.

Conversion and Activation Review

Review the journey from landing page and lead capture through registration, first value and transparent upgrades.

Campaign Content and Brand Review

Plan a campaign, draft its channel content and review the work against actual brand guidance.

Content and Launch Planning

Prioritize an editorial roadmap and plan how to launch and distribute it across suitable channels.

Customer, Market and Community Research

Understand customer needs, compare competitors and plan a community around real member value.

Lead Magnet to Lifecycle Nurture

Choose a relevant lead magnet, then draft a permission-based nurture journey with entry, suppression and exit rules.

API Contract and Observability

Define the API contract, then plan how to observe its latency, failures and retries. Guidance and checklist; no production changes.

Data Analysis Foundation

Profile a dataset, choose and interpret statistical methods, then validate calculations and conclusions before sharing.

Choose an area

01 · Category

What the MCP returns at this step

87 · Marketing · Creative Research

Competitor Ad Intelligence

Summarize sampled competitor ads into cross-brand patterns and whitespace. Use Competitor Ad Collection for detailed cohort collection.

wettbewerbsanzeigenberichtcompetitor-ad-intelligence

Competitor Ad Intelligence · Original SKILL.md

---
name: competitor-ad-intelligence
description: Analyze live competitor ads into a cross-brand creative teardown — angles, hooks, formats, investment proxies, and sampled market whitespace. Use when the user wants a summarized competitor report or briefing input. For detailed competitor discovery and evergreen-versus-breakout cohort collection, use creative-research. Requires GoMarble MCP; file inputs cannot supply live competitor data.
---

# Competitor Ad Intelligence

Turns GoMarble's Ads Library MCP tools into a structured competitive-creative brief: what competitors are running right now, what patterns repeat across their prominent ads, and what that implies for your own account.

## When to use this skill

- "Turn these competitor ads into a cross-brand creative teardown"
- "What patterns and investment signals repeat across these competitors?"
- "Summarize this competitor research into briefing inputs and sampled market whitespace"
- Before any new creative brief — as the competitive-context input into `winning-pattern-synthesis`

For raw discovery or collection requests such as "find competitors," "show me their ads," or "research the ad landscape," use `creative-research` first, then apply this skill if the user also wants a synthesized teardown.

## Hard requirement

This skill only works through GoMarble MCP. If GoMarble MCP is not connected in the current session, say so explicitly and stop — do not attempt to approximate this with web search or general knowledge of a brand's marketing. Competitor ad library data is not something Claude can infer or recall reliably; the whole value is that it's live.

Within the Marketing Agent plugin, `creative-research` owns detailed discovery and evergreen/breakout cohort collection. Use this skill to interpret those results across brands and produce the structured competitive brief; call the tools below directly only when the needed ads have not already been collected.

## Tools (GoMarble MCP `ads_library_*`)

| Tool | Use for |
|---|---|
| `ads_library_search_brands` | Resolve a brand name to a brand ID before pulling ads |
| `ads_library_find_competitors` | AI + web-search assisted competitor discovery when the user gives a brand but not a competitor list |
| `ads_library_get_ads_by_brand_id` | Pull a brand's ad set with filtering/pagination |
| `ads_library_discover_ads` | Keyword/filter-based ad discovery across the whole database (category research, not brand-specific) |
| `ads_library_analyze_ad` | Deep single/multi-ad analysis — creative breakdown + duplicate-variation detection |
| `ads_library_get_brand_analytics` | Brand-level spend/volume/velocity stats |

## Workflow

### Step 1: Establish scope
Confirm with context already available (don't ask if it's inferable):
- Target brand(s) — the user's own account, or the competitor(s) named
- Category, if no specific competitor is named yet
- Platform(s) — Meta, Google, or both
- What they actually want: a landscape scan, a deep-dive on 1-2 named competitors, or ongoing tracking

### Step 2: Resolve brands
- If competitors are named: `ads_library_search_brands` to resolve each to a brand ID
- If competitors are not named: `ads_library_find_competitors` using the user's brand/category — this does its own research, so don't pre-guess competitors from training data
- If it's a category scan with no specific brand: skip straight to `ads_library_discover_ads` with category/keyword filters

### Step 3: Pull the ads
- Per competitor: `ads_library_get_ads_by_brand_id` — pull a meaningful sample (aim for enough to see repeated patterns, not just the newest 2-3 ads)
- Cross-check spend/scale with `ads_library_get_brand_analytics` — high spend or long duration is evidence that the brand continues to invest in an ad, not proof that it performs well. Duration and apparent scale are prioritization proxies because Claude has no access to the competitor's actual outcome metrics.

### Step 4: Analyze the creative
For the ads worth analyzing in depth (longest-running, highest apparent spend, or most-duplicated), run `ads_library_analyze_ad`:
- Extract: hook (first 3 seconds / first line), core angle, format, offer/CTA, visual style
- `ads_library_analyze_ad` also surfaces duplicate variations of the same ad. Many variants can indicate continued investment or structured testing, but not a confirmed winner; note the variation count without assigning performance the data does not show.

### Step 5: Pattern-tag every ad reviewed
Don't just describe ads one by one — tag each with:
- **Angle** (problem-agitation, social proof, comparison, founder-story, before/after, price/offer-led, etc.)
- **Hook mechanism** (question, bold claim, pattern interrupt, UGC-style testimonial opener, etc.)
- **Format** (single video, single image, carousel, catalog)
- **Signal strength** (running long / many variants = stronger investment proxy; new/single variant = limited signal)

## Output Format

```
### Competitor Ad Intelligence: [Category/Brand(s)]

Scope: [brands reviewed, platform(s), date pulled]

Per-competitor summary:
| Brand | Ads reviewed | Dominant angle | Dominant format | Scale signal |
|---|---|---|---|---|

Cross-competitor patterns (repeat across 2+ brands):
- [Pattern]: seen in [brands]; priority rationale: [duration/spend/variation proxy evidence]

Notable single-brand standouts:
- [Brand]: [what's unusual/aggressive about their approach]

Whitespace (patterns competitors aren't using):
- [gap the user could test]

Caveats:
- This reflects what's live/visible in the ads library, not verified performance data — treat scale/duration as a proxy signal, not ground truth.
```

## Guardrails

- Never present a competitor's ad copy or creative as something to copy verbatim — this is pattern/angle intelligence for inspiration, not a rip-and-replace source. Feed patterns into `winning-pattern-synthesis` and `ad-brief-generator` for original creative.
- Don't overstate confidence — "running for 40 days with 6 variants" is a real signal; a single ad seen once is not evidence of anything.
- If `ads_library_find_competitors` returns brands the user doesn't recognize as real competitors, flag that rather than silently including them.

When to use

Summarize sampled competitor ads into cross-brand patterns and whitespace. Use Competitor Ad Collection for detailed cohort collection.

What you get

The complete original GoMarble guidance, with source attribution and declared limits.

How it works

Clarify scope, use available evidence, and apply the relevant source workflow within actual user authorization.

Requirements

Competitor brands, geography, observation window and GoMarble live ad-library access.

Delivery and review notes

Only sampled accessible ads support conclusions; live competitor data cannot be invented from files. Investment proxies are not measured spend or return. GoMarble connectors, referenced sibling skills and runtime dependencies are not bundled or installed by M11. Brief obvious-danger screening only; no functional test or comprehensive safety certification. Loading this text authorizes no external calls, account mutations or ad spend.

Starting prompt

Use Competitor Ad Intelligence for [TASK]. Clarify Competitor brands, geography, observation window and GoMarble live ad-library access. Separate observed evidence from heuristics and recommendations. Check actual connector availability and approval semantics before any external action.

Not for

Claiming installed GoMarble tools, proven campaign outcomes or authorization to launch ads merely from loading this source.

What M11 added

German routing and explicit scope. Original attribution: GoMarble. Only sampled accessible ads support conclusions; live competitor data cannot be invented from files. Investment proxies are not measured spend or return. GoMarble connectors, referenced sibling skills and runtime dependencies are not bundled or installed by M11. Brief obvious-danger screening only; no functional test or comprehensive safety certification. Loading this text authorizes no external calls, account mutations or ad spend.

Original authorship remains with gomarble-ai/marketing-agent · Original source ↗

Original license & copyright

MIT License

Copyright (c) 2026 GoMarble

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Source SHA-256: 4cb37bf020c0df4c0473c110029302ee3dbc17a276bc61a3003ad9b38eb97aff
Snapshot checked: 2026-09-28T16:19:52.867Z

Find a skill.Get to work.

All skills

Multi-Agent Orchestration

AI Context Window Audit

AI Agent Architecture Audit

AI Skill Discovery

Lead Magnet Strategy

Ideal Customer Profile

Go-to-Market Strategy

Marketing Campaign Ideas

Product Growth Loops

Competitor Analysis

North Star Metric

Product Positioning

Product Vision

Go-to-Market Motion Selection

Customer Feedback and JTBD Analysis

PESTLE Market Environment Analysis

Customer Journey Mapping

Ansoff Growth Options

Data Analysis Validation

Dataset Profiling

Statistical Analysis Guidance

Programmatic SEO Planning

Landing Page and Form Conversion Review

Paywall and Upgrade Planning

Signup and Registration Review

User Onboarding and Activation

Popup and Modal Planning

Email Sequence Copy and Flow

Marketing Campaign Planning

Marketing Content Drafting

Brand Voice and Content Review

Marketing Performance Reporting

Sales Company Research

Company and Contact Enrichment

Website Information Architecture

Content Strategy and Editorial Roadmap

Product Launch Planning

Customer Research and Voice of Customer

Community Growth and Member Experience

Competitor Research Profiles

Roadmap and Release Communication

Lifecycle Email Sequences

API Contracts and Interface Design

Observability and Instrumentation Planning

Agent Context and Session Handoff

Product Discovery Sprint

Deal Quality Scoring

Enrichment Waterfall Design

Customer Retention Playbook

Sales Coaching Practice

Revenue Cohort Analysis

Intent Signal Scoring

Customer Identity Matching

Segment Activation Planning

Sales Call Review

Retention Dashboard Design

Sales Coaching Competencies

Offer Delivery Design

Offer Value Communication

Service Productization

Conversion Hypothesis Prioritization

Braze Engagement Guidance

Brand Voice Guide Design

Content Pipeline Orchestration

Website Brand Profile

Buyer Objection Clarification

Klaviyo Integration Guidance

Offer Bonus Planning

Klaviyo Marketing Review

Cross-Platform Advertising Review

Expertise Business Model

LinkedIn Advertising Guidance

TikTok Advertising Guidance

Evidence Research Brief

Cinematic Video Prompts

Visual Prompt Model Adaptation

Visual Prompt Diagnostics

Short Film Development

Find a skill.
Get to work.