A community-driven registry for Claude, Cursor, Windsurf, Cline & more. Not affiliated with Anthropic.
Are you the author? Sign in to claim
Meta Ads Analyzer skill + MCP server for Claude Code. Breakdown Effect, Learning Phase, and expert-level campaign diagno
A Claude Code skill + MCP server setup for expert-level Meta Ads campaign analysis. Includes the Breakdown Effect framework, Learning Phase diagnostics, and 9 reference documents from Meta's official documentation.
When installed, Claude Code can:
| Component | What it does |
|---|---|
Skill (skill/) | Analysis framework with 9 reference docs that Claude loads as context |
MCP Server (mcp/) | Connects Claude Code to Meta's Marketing API for live data |
Scripts (scripts/) | Setup and token refresh helpers |
Copy the skill/ folder into your Claude Code project:
# From your Claude Code project root
mkdir -p .claude/skills/meta-ads-analyzer
cp -r skill/* .claude/skills/meta-ads-analyzer/
The skill is now active. Claude will automatically use it when you ask about Meta Ads analysis.
The MCP server lets Claude pull live campaign data from Meta's API. Skip this if you only want to analyze exported data (CSV, screenshots).
ads_read, ads_management, business_managementbash scripts/setup.sh
Copy the MCP config template and add your credentials:
cp mcp/mcp.json.example .mcp.json
Edit .mcp.json and replace the placeholders with your actual token and app secret.
Then restart Claude Code for the MCP server to connect.
Ask Claude: "Analyze my Meta Ads campaigns" — it should connect to your ad account and start analyzing.
Meta long-lived tokens expire after ~60 days. To refresh:
bash scripts/refresh_token.sh
If the token already expired, generate a new one from the Graph API Explorer and run scripts/setup.sh again.
Once installed, you can ask Claude things like:
You can also paste CSV exports or screenshots from Meta Ads Manager — the skill works with any data format.
| Document | Covers |
|---|---|
breakdown_effect.md | Why Meta allocates budget to "expensive" segments |
core_concepts.md | Ad Auction, Pacing, Learning Phase overview |
learning_phase.md | How learning works, when it resets |
ad_relevance_diagnostics.md | Quality, Engagement, Conversion rankings |
auction_overlap.md | When your ads compete against each other |
pacing.md | Budget and bid pacing mechanics |
bid_strategies.md | Spend-based, goal-based, manual bidding |
ad_auctions.md | How auction winners are determined |
performance_fluctuations.md | Normal vs. concerning performance changes |
MIT
A Jetbrains IDE IntelliJ plugin aimed to provide coding agents the ability to leverage intelliJ's indexing of the codeba
Run Claude Code as an MCP server so any agent can delegate coding tasks to it
Browser automation using accessibility snapshots instead of screenshots