Growth Nirvana Skills

Give your AI assistant deep, project-aware knowledge of Growth Nirvana bundles, model variants, and MCP query workflows.

Install

Run from a Cursor project:

npx @growthnirvana/skills add

Install globally (all projects):

npx @growthnirvana/skills add --global

Create a project mcp.json template:

npx @growthnirvana/skills init-mcp

You can also run the installed binary directly:

gn-skills add
gn-skills add --global
gn-skills init-mcp

MCP Server Setup

Installing MCP config per project is normal. Use:

gn-skills init-mcp

This creates .cursor/mcp.json in your current repository. It uses environment variables for credentials and launches growth-nirvana-mcp-server via npx:

"/bin/zsh -lc \"... if [ -z \\\"$GROWTH_NIRVANA_API_KEY\\\" ]; then ...; npx -y growth-nirvana-mcp-server\""

Set your key in either your project .env.local (recommended) or shell profile:

echo 'GROWTH_NIRVANA_API_KEY="your_api_key_here"' >> .env.local

Or shell profile (~/.zshrc, ~/.bashrc, etc.):

export GROWTH_NIRVANA_API_KEY="your_api_key_here"

Then reload Cursor.

Useful options:

• gn-skills init-mcp --target .cursor/mcp.json
• gn-skills init-mcp --force
• gn-skills init-mcp --pin-server-version 1.2.3

Verify env var expansion

1. Confirm .env.local expansion:

echo 'GROWTH_NIRVANA_API_KEY="test_key_123"' > .env.local
/bin/zsh -lc 'set -a; [ -f .env.local ] && source .env.local; set +a; echo "$GROWTH_NIRVANA_API_KEY"'

2. Confirm generated config has shell-based launch:

node scripts/install.js init-mcp --target .tmp-install-test/mcp.json --force

3. In Cursor, reload and start the MCP server. If the key is missing, startup fails with:

Missing GROWTH_NIRVANA_API_KEY

What It Enables

• Bundle export workflows
• Query answering from downloaded bundles
• Variant-aware model selection (combined_report, combined_report_ff, combined_report_catalog)
• MCP dry-run/execute flows with result retrieval (include=results)

Example Prompts

• "What channels is <client name> using?"
• "Spend by channel for Feb 2026 for <client name>."
• "For Deepgram, give me CTR and CPC by channel for Feb 2026."
• "Draft a new transformation and dry-run it."

How It Works

1. Detect dataset via dataset/dataset.json and displayName.
2. Select the correct reporting model variant.
3. Validate columns from warehouse field metadata.
4. Execute via MCP when needed.
5. Return evidence-backed answers.

Included Skills

• growth-nirvana-dataset-export
• growth-nirvana-bundle-query-transform
• growth-nirvana-bundle-query
• growth-nirvana-transform-authoring

Development

From this repo root:

node scripts/install.js add
node scripts/install.js add --global
node scripts/install.js init-mcp

Publishing To npm

1. Login:

npm login

2. Verify package contents:

npm pack --dry-run

3. Publish:

npm publish --access public

4. Validate install:

npx @growthnirvana/skills add
npx @growthnirvana/skills init-mcp