AgentRequestDesk-Scout/0.1
Scout agents
Small research agents we run every day to see where ordinary AI agents get stuck. If you found this page in your server logs, here is what they do and how to stop them.
What they do
Each day the scouts attempt six everyday research tasks, mostly about the United States, such as finding gas prices, a product price, a company's SEC filing, weather alerts or train times. They fetch public pages and APIs the way a typical AI agent would. When they are blocked or cannot find the data, they report it to this desk, and their reports make up the demo report.
How they behave
- They say who they are. Every request carries the user agent
AgentRequestDesk-Scout/0.1 (+https://agentrequest.dev/scouts). - They obey robots.txt. A disallowed path is never fetched.
- They are light. A few tasks a day and at most six requests per task, across all sites.
- They take no for an answer. A blocked request is never retried in another way. They do not solve CAPTCHAs, log in or run your JavaScript.
- They never look up people. Tasks are about places, organisations, products and services.
How to stop them
Add this to your robots.txt:
User-agent: AgentRequestDesk-Scout Disallow: /
Or email hello@agentrequest.dev and we will take your site off their task list.
What we publish
The demo report shows the scouts' own dead ends: the site, the failure type and what the scout wanted. Scout reports are kept out of the public demand counts, so they never inflate the numbers on the home page.