A tool to analyze how "agentic" your AI product really is. Uses GPT-4 and Claude to provide honest scores on autonomy, goal-oriented behavior, persistence, and adaptability.
Tired of every AI product calling itself an "agent"? This tool helps you analyze how truly agentic your AI product idea is by:
- Scoring key agent characteristics (autonomy, goal-oriented behavior, persistence, adaptability)
- Getting brutally honest feedback from both GPT-4 and Claude
- Providing detailed explanations for each score
- Showing an overall "agenticity" score
- Cloudflare account
- OpenAI API key (GPT-4 access required)
- Anthropic API key (Claude access required)
- Node.js and npm installed
- Wrangler CLI installed (
npm install -g wrangler
)
-
Clone the repository:
git clone https://github.com/yourusername/agent-or-not.git cd agent-or-not
-
Create an R2 bucket in Cloudflare dashboard named
agent-or-not
-
Copy wrangler.toml.example to wrangler.toml:
cp wrangler.toml.example wrangler.toml
-
Update the API keys in wrangler.toml:
[vars] OPENAI_API_KEY = "your-openai-api-key" CLAUDE_API_KEY = "your-claude-api-key"
-
Login to Cloudflare using Wrangler:
wrangler login
-
Deploy the static assets to R2:
npm run deploy:html
-
Deploy the worker:
npm run deploy:worker
-
Run local development server:
npm run dev
-
The server will be available at
http://localhost:8787
-
Add your domain in Cloudflare dashboard
-
Update wrangler.toml with your domain:
[[routes]] pattern = "yourdomain.com" custom_domain = true
-
Deploy again:
npm run deploy
Required environment variables in wrangler.toml:
OPENAI_API_KEY
: Your OpenAI API key with GPT-4 accessCLAUDE_API_KEY
: Your Anthropic API key for Claude access
MIT License
Shyjal