The open-source one-stop shop for advertising platform APIs, MCP servers, AI agents, and cross-platform automation. 14 platforms. 25+ specialized agents. Production tooling. Built by practitioners who manage real ad spend.

Connected Properties: itallstartedwithaidea.com · googleadsagent.ai

Every ad platform has scattered repos — SDKs in 6 languages, deprecated libraries, mobile examples no media buyer will ever touch. Nobody has stitched them together.

The Advertising Hub does three things no single platform repo does:

1. Curates the advertising-relevant repos from 14 platforms (filtered — no mobile SDKs, no deprecated SOAP libraries, only what practitioners need)
2. Connects them with a shared core package, specialized AI agents, and MCP servers that give AI tools live API access
3. Documents the cross-platform patterns, gotchas, and strategies that come from 15+ years of managing enterprise ad spend

Legend: ✅ Live = Production MCP server available · 📋 Spec Ready = Architecture documented, ready to build · 📋 Planned = Platform module exists with patterns

From The Agency Enhanced — 7 agents with production tooling from googleadsagent.ai:

MCP (Model Context Protocol) servers give AI tools like Claude, Cursor, and Gemini direct access to advertising platform APIs.

Want to build an MCP server? Start with our templates and development guide.

The core/ package provides shared utilities that work across all 14 platforms:

• core/auth/ — Unified OAuth2/API key patterns for every platform
• core/models/ — Normalized campaign, ad group, ad, audience, and metrics models
• core/rate_limiting/ — Platform-specific adaptive rate limiting
• core/errors/ — Unified error taxonomy mapping platform errors to common types
• core/utils/ — Currency handling, date ranges, pagination, response normalization

Tools built and deployed from this ecosystem:

# Copy agents to your Claude Code directory cp -r agents/ ~/.claude/agents/advertising-hub/ # Activate in Claude Code: # "Use the PPC Campaign Strategist agent to audit this account" # "Use the Budget Allocator agent to recommend cross-platform spend"

# Generate integration files for Cursor, Gemini CLI, Windsurf, etc. ./scripts/convert.sh # Interactive install (auto-detects your tools) ./scripts/install.sh

from core.auth.google import GoogleAdsAuth from core.auth.meta import MetaAdsAuth from core.models.campaign import Campaign from core.models.metrics import NormalizedMetrics # Unified auth across platforms google_auth = GoogleAdsAuth(client_id="...", client_secret="...", refresh_token="...") meta_auth = MetaAdsAuth(access_token="...", app_id="...", app_secret="...") # Normalized data models work with any platform campaign = Campaign( platform="google-ads", name="Brand - US", budget_daily=500.00, status="active" )

The wiki is the knowledge base — practitioner-written documentation for every platform, pattern, and tool:

• Platform Guides — Deep dives into each platform's API, auth, and gotchas
• Cross-Platform Patterns — Auth, tracking, audiences, reporting compared
• MCP Development Guide — How to build MCP servers for ad platforms
• Agent Guide — How to use, customize, and build agents
• Use Cases — Real-world scenarios with step-by-step playbooks

advertising-hub/ ├── README.md # You are here ├── core/ # 🔧 Shared utilities across all platforms │ ├── auth/ # Unified auth (OAuth2, API keys) per platform │ ├── models/ # Normalized campaign/ad/audience/metrics models │ ├── rate_limiting/ # Adaptive rate limiting per platform │ ├── errors/ # Unified error taxonomy │ └── utils/ # Currency, dates, pagination, normalization ├── platforms/ # 📡 Platform modules (14 platforms) │ ├── google-ads/ # Agents, scripts, patterns, MCP, upstream refs │ ├── meta-ads/ │ ├── microsoft-ads/ │ ├── amazon-ads/ │ ├── linkedin-ads/ │ ├── pinterest-ads/ │ ├── reddit-ads/ │ ├── spotify-ads/ │ ├── thetradedesk/ │ ├── demandbase/ │ ├── criteo/ │ ├── adroll/ │ └── quora-ads/ ├── agents/ # 🤖 AI agent specs (25+) │ ├── paid-media/ # Battle-tested paid media agents │ ├── cross-platform/ # Multi-platform coordination agents │ ├── platform-specific/ # Platform-native specialists │ └── orchestrator/ # Buddy + meta-agents ├── mcp-servers/ # 🔌 MCP server registry + templates ├── wiki/ # 📖 Complete documentation ├── integrations/ # 🔗 IDE integrations (Claude Code, Cursor, etc.) ├── scripts/ # 🛠️ Build, install, lint scripts ├── .github/ # CI/CD workflows └── examples/ # 📚 Real-world playbooks 

We welcome contributions! The most impactful ways to help:

1. Build an MCP server for a platform that doesn't have one yet — start with templates
2. Add a platform specialist agent — follow the agent template
3. Write a pattern doc — share hard-won knowledge about a platform's API gotchas
4. Improve cross-platform guides — help connect the dots between platforms

See CONTRIBUTING.md for full details.

MIT License — Use freely, commercially or personally.

Built by John Williams — Senior Paid Media Specialist at Seer Interactive with 15+ years managing enterprise ad spend across Google, Meta, Microsoft, and Amazon. Builder of googleadsagent.ai. Speaker at Hero Conf on AI applications in advertising.

Agent Foundation: The Agency by Michael Sitarzewski — MIT Licensed.