Files
our-claude-skills/ga-agent-project/01-mcp-setup/README.md
Andrew Yim 214247ace2 feat(ga-agent): Add GA Agent project with decomposed architecture
Create workspace for building Google Analytics Claude Skill with:
- 5 independent components (MCP setup, skill, dimension explorer, Slack reporter, realtime watcher)
- Comprehensive project plan and documentation
- Step-by-step setup guides for each component

Components:
1. MCP Setup - GA4 + BigQuery MCP server installation
2. GA Agent Skill - Core Claude Skill for interactive analysis
3. Dimension Explorer - Validate dims/metrics with explanations
4. Slack Reporter - Automated reports to Slack (P2)
5. Realtime Watcher - Real-time monitoring (deferred)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-21 10:32:13 +09:00

144 lines
3.0 KiB
Markdown

# Component 1: MCP Setup
**Type:** Infrastructure
**Priority:** P0
**Status:** Not Started
## Goal
Install and configure GA4 + BigQuery MCP servers for Claude Code.
## Prerequisites
- Google Cloud account
- GA4 property access
- `gcloud` CLI installed
## Setup Steps
### Step 1: Google Cloud Project
```bash
# Authenticate
gcloud auth login
# Set project (create if needed)
gcloud config set project YOUR_PROJECT_ID
# Enable APIs
gcloud services enable \
analyticsdata.googleapis.com \
analyticsadmin.googleapis.com \
bigquery.googleapis.com
```
### Step 2: Service Account
```bash
# Create service account
gcloud iam service-accounts create ga-mcp-agent \
--display-name="GA MCP Agent"
# Get email
SA_EMAIL="ga-mcp-agent@YOUR_PROJECT_ID.iam.gserviceaccount.com"
# Grant BigQuery roles
gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
--member="serviceAccount:$SA_EMAIL" \
--role="roles/bigquery.dataViewer"
gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
--member="serviceAccount:$SA_EMAIL" \
--role="roles/bigquery.jobUser"
# Download key
gcloud iam service-accounts keys create \
../config/service-account.json \
--iam-account=$SA_EMAIL
```
### Step 3: GA4 Property Access
1. Go to [GA4 Admin](https://analytics.google.com/)
2. Property → Property Access Management
3. Add user: `ga-mcp-agent@YOUR_PROJECT_ID.iam.gserviceaccount.com`
4. Role: Viewer
### Step 4: Install GA4 MCP Server
```bash
# Clone official server
git clone https://github.com/googleanalytics/google-analytics-mcp.git
cd google-analytics-mcp
# Setup Python environment
python -m venv venv
source venv/bin/activate
pip install -e .
# Test
export GOOGLE_APPLICATION_CREDENTIALS="../config/service-account.json"
python -m google_analytics_mcp --help
```
### Step 5: Install BigQuery MCP Server
```bash
# Test with npx (no install needed)
npx -y @ergut/mcp-bigquery-server \
--project-id YOUR_PROJECT_ID \
--location us-central1 \
--key-file ../config/service-account.json
```
### Step 6: Configure Claude Code
Add to `~/.claude/mcp_servers.json`:
```json
{
"mcpServers": {
"google-analytics": {
"command": "python",
"args": ["-m", "google_analytics_mcp"],
"cwd": "/path/to/google-analytics-mcp",
"env": {
"GOOGLE_APPLICATION_CREDENTIALS": "/path/to/service-account.json"
}
},
"bigquery": {
"command": "npx",
"args": [
"-y",
"@ergut/mcp-bigquery-server",
"--project-id", "YOUR_PROJECT_ID",
"--location", "us-central1",
"--key-file", "/path/to/service-account.json"
]
}
}
}
```
### Step 7: Verify
```bash
# Restart Claude Code, then:
mcp-cli servers
# Should show: google-analytics, bigquery
mcp-cli tools google-analytics
mcp-cli tools bigquery
```
## Checklist
- [ ] GCP project configured
- [ ] APIs enabled
- [ ] Service account created
- [ ] GA4 access granted
- [ ] GA4 MCP installed
- [ ] BigQuery MCP installed
- [ ] Claude Code configured
- [ ] Connection verified