$ cd ~/gtm-os/mcp/

The MCP Stack

Model Context Protocol is how AI agents connect to tools, data sources, and APIs. This is the MCP infrastructure powering theGTMOS.ai - from web research to browser automation to content generation.

5MCP Servers

3Agent Types

4Cron Jobs

$ ls ~/mcp/servers/

Configured Servers

ExaACTIVE

Web search, company research, and code context. Powers prospect intelligence, competitive analysis, and real-time market data.

›Company research enrichment
›Technical documentation lookup
›Market trend analysis

PlaywrightACTIVE

Browser automation and visual QA. Headless browser control for testing, screenshots, and form interactions.

›Landing page visual QA
›Automated screenshot generation
›Form testing and validation

FilesystemACTIVE

Direct file system access for reading, writing, and managing project files. Powers content generation and data pipeline outputs.

›Blog post generation to disk
›SEO data file management
›Configuration updates

GitHubACTIVE

Repository management, PR creation, issue tracking, and code review automation via GitHub CLI integration.

›Automated PR creation
›Issue triage and labeling
›Code review assistance

Custom ToolchainIN DEV

Specialized tools built for GTM workflows - Clay enrichment connectors, CRM sync utilities, and email infrastructure management.

›Clay table automation
›HubSpot data sync
›Sending infrastructure audits

$ ls ~/mcp/agents/

Agent Patterns

Parallel agents handle complex multi-step tasks. Each agent type specializes in a domain and has access to specific MCP servers.

Research Agents

Exa web search + company intelligence. Parallel research across multiple prospects, competitors, or market segments.

ExaFilesystem

QA Agents

Playwright browser automation. Visual testing, screenshot capture, and cross-page validation.

PlaywrightFilesystem

Content Agents

Blog generation, SEO optimization, and content pipeline management. Combines research with writing.

ExaFilesystemGitHub

$ agent-dispatch --parallel|-- research-agent -> Exa.company_research("clay.com")|-- research-agent -> Exa.web_search("GTM automation 2026")|-- qa-agent -> Playwright.screenshot("/features")`-- content-agent -> Generate blog draft from research[4 agents] [parallel] [avg 12s completion]

$ crontab -l

Always-On Automation

Background processes that run daily via launchd. No manual intervention required.

# Daily blog generation0 6 * * * python3 nio_blog_generator.py # STATUS: ACTIVE

# SEO keyword pipeline0 7 * * * python3 seo_keyword_pipeline.py # STATUS: ACTIVE

# Daily tracker scan0 8 * * * python3 daily_tracker_scan.py # STATUS: ACTIVE

# Content index rebuild0 9 * * 1 rebuild-content-index # STATUS: WEEKLY

$ cat ~/mcp/integration-map.txt

The GTM Data Flow

How MCP connects the entire GTM stack - from data enrichment to outreach.

Clay Enrichment

Contacts

→

Exa Research

Intel Data

→

Claude Analysis

Scored Leads

→

Instantly Outreach

Email Sends

→

HubSpot CRM

Pipeline

← home system vitals →