$Compounding Marketing

MCP Integrations

Enhance skills with Perplexity and Exa for real-time research and competitor intelligence

MCP Integrations

Compounding Marketing integrates with Model Context Protocol (MCP) servers to unlock real-time research and deep competitive intelligence.

What is MCP?

MCP (Model Context Protocol) is an open standard for connecting AI models to external tools and data sources. Think of it as "plugins for AI."

Without MCP: Skills work with static data (your .cm-context file, past deliverables/)

With MCP: Skills can:

  • Pull live web research (Perplexity)
  • Find competitor data and people profiles (Exa)
  • Access your CRM, analytics, calendar in real-time

Supported MCPs

| MCP | What It Does | Which Skills Use It | |-----|-------------|---------------------| | Perplexity | Real-time web research, news, trends | Research, positioning, content creation | | Exa | Competitor research, people search, company data | Competitive analysis, ICP research, outreach |

Perplexity MCP

What It Unlocks

Perplexity gives skills real-time web knowledge:

  • Current trends and news
  • Competitor analysis
  • Market research
  • Technical documentation

Example: Without Perplexity

You run /cm:research:

Based on your context, here are trending topics in personal branding:
1. AI-generated content
2. LinkedIn growth strategies
3. Founder-led marketing

Generic. No citations. Might be outdated.

Example: With Perplexity

You run /cm:research:

Based on live web research (last 7 days), here are trending topics in personal branding for CTOs:

1. **"How CTOs are using AI to build in public"**
   - 3 viral posts this week (500+ likes each)
   - Key angle: Transparency + AI demos
   - Sources: [LinkedIn posts, Hacker News discussion]

2. **"Technical founder personal brand playbooks"**
   - New post from [Founder Name] got 10k views
   - Key takeaway: "Show your thinking, not just results"
   - Sources: [Blog post URL, Twitter thread]

3. **"AI-native product development"**
   - Trending on Twitter (2.5k mentions this week)
   - Key pain: "Everyone says 'AI-native' but what does it mean?"
   - Opportunity: Explainer content

Real data. Recent. Cited.

How to Install

Claude Code / Claude Desktop

claude mcp add perplexity -- npx -y @perplexity-ai/mcp-server

Then add your Perplexity API key to the config:

macOS:

code ~/Library/Application\ Support/Claude/claude_desktop_config.json

Windows:

%APPDATA%\Claude\claude_desktop_config.json

Add:

{
  "mcpServers": {
    "perplexity": {
      "command": "npx",
      "args": ["-y", "@perplexity-ai/mcp-server"],
      "env": {
        "PERPLEXITY_API_KEY": "your_api_key_here"
      }
    }
  }
}

Get your API key: perplexity.ai/settings/api

OpenClaw

npx mcporter add perplexity

Then edit config/mcporter.json to add your API key.

Which Skills Use Perplexity

| Skill | How It Uses Perplexity | |-------|------------------------| | /cm:research | Finds trending topics, competitor content | | /cm:icp-research | Pulls latest pain points, jobs-to-be-done data | | /cm:competitive-analysis | Gets real-time competitor updates | | /cm:content-brief | Researches top-ranking content for keywords | | /cm:position | Validates category fit, checks market trends |


Exa MCP

What It Unlocks

Exa gives skills deep search and structured data:

  • Company profiles (employees, tech stack, funding)
  • People search (find executives, decision-makers)
  • Competitor tracking
  • Semantic search (find pages by meaning, not just keywords)

Example: Without Exa

You run /cm:icp-research:

Your ICP is: CTOs at B2B SaaS companies.

Typical pain points:
- Hiring challenges
- Managing remote teams
- Building in public

Generic. No real data.

Example: With Exa

You run /cm:icp-research:

Your ICP is: CTOs at B2B SaaS companies (50-200 employees).

Live data from Exa:

**Companies matching your ICP:**
1. **Acme Corp** (Series A, 75 employees, tech stack: React, Python, AWS)
   - CTO: Jane Doe (LinkedIn: [URL])
   - Recent funding: $5M
   - Pain signals: Hiring post for "Head of Eng" (posted 3 days ago)

2. **BetaCo** (Series B, 120 employees)
   - CTO: John Smith
   - Recent blog post: "Why we're struggling with remote onboarding"
   - Pain signal: Looking for tools to scale team culture

**Outreach angle:**
"I saw you're hiring. Here's how 3 CTOs at similar-stage companies solved remote onboarding with [your tool]."

Sources: Exa company search + LinkedIn + recent blog posts

Real companies. Real pain signals. Actionable.

How to Install

Claude Code / Claude Desktop

claude mcp add exa -- npx -y @exa-labs/mcp-server

Add your Exa API key:

{
  "mcpServers": {
    "exa": {
      "command": "npx",
      "args": ["-y", "@exa-labs/mcp-server"],
      "env": {
        "EXA_API_KEY": "your_api_key_here"
      }
    }
  }
}

Get your API key: exa.ai/api

OpenClaw

npx mcporter add exa

Edit config/mcporter.json to add your API key.

Which Skills Use Exa

| Skill | How It Uses Exa | |-------|-----------------| | /cm:icp-research | Finds real companies matching your ICP | | /cm:competitive-analysis | Pulls competitor tech stack, team size, funding | | /cm:cold-email | Finds decision-makers and pain signals for outreach | | /cm:position | Researches companies in your category |


How MCPs Enhance Specific Skills

/cm:research + Perplexity

Before: Generic trending topics

After: Real trending topics from the last 7 days with citations

Use case: Daily content creation. Find what's actually trending, not what ChatGPT thinks is trending.


/cm:position + Exa

Before: "Based on your context, here's your positioning"

After: "Here are 5 companies in your category. Here's how they position. Here's where you can differentiate."

Use case: Validate category fit. Make sure you're not positioning into a dead market.


/cm:icp-research + Exa

Before: Generic persona ("CTOs at B2B SaaS")

After: 10 real companies with real CTOs, their tech stack, recent hires, blog posts

Use case: Build a real outreach list, not a hypothetical persona.


/cm:cold-email + Perplexity + Exa

Before: Generic cold email template

After: Personalized email with:

  • Their recent blog post (Perplexity)
  • Their company's tech stack (Exa)
  • Pain signals (hiring posts, tweets)

Use case: Write cold emails that reference real, recent context. Massively higher reply rates.


Troubleshooting MCP Connections

"Perplexity not responding"

  1. Check API key is correct
  2. Verify you have credits (free tier: 5 requests/day)
  3. Restart Claude Desktop completely

"Exa search returns no results"

  1. Try broader search terms (e.g., "B2B SaaS" instead of "AuthorityMax competitors")
  2. Check API key permissions
  3. Verify you have credits

"Skills aren't using MCPs"

MCPs are optional. Skills will run without them, just with less real-time data.

To force a skill to use an MCP:

/cm:research --use-perplexity

Or update .cm-context:

# Research Settings
Use Perplexity: Yes
Use Exa: Yes

Do You Need MCPs?

You DON'T need them if:

  • You're just starting (focus on positioning first)
  • You have strong customer research already
  • You're in a well-defined market (no need for validation)

You DO need them if:

  • You're doing daily content creation (need trending topics)
  • You're building outreach lists (need real companies/people)
  • You're validating a new category (need market data)
  • You're analyzing competitors regularly

Start without MCPs. Add them when you need real-time data.


Next Steps