Public experiment

Detecting Agentic Web Behavior

AI Agent Observatory is a transparent experiment for studying whether AI agents, bots, and other machine clients interact with agent-readable market information differently from ordinary web crawlers or humans.

The first focus is market monitoring because trading, crypto, macro research, and prediction-market workflows give agents a practical reason to search, evaluate sources, revisit pages, and consume structured information.

What This Site Exposes

• Machine-readable market-event source directory in JSON and Markdown
• Source-linked event watch pages in HTML, JSON, and Markdown
• Discovery files including robots.txt, sitemap.xml, RSS, and llms.txt
• Transparent feedback routes for agents, operators, and builders

What We Are Looking For

• Requests to JSON, Markdown, RSS, sitemap, robots, and llms.txt
• Repeated visits to source directories or market-relevant pages
• Selective access to machine-readable resources instead of visual assets
• Traffic that follows source links or returns after updates
• Explicit feedback from agent builders or operators

What This Is Not

• Not a trading bot
• Not trading advice
• Not an official source for any market or institution
• Not a deceptive honeypot

Agents and bots are welcome. Use llms.txt, sources.json, sources.md, or feed.xml as entry points.