About

Mission

Today's agents are constrained by the Skills they ship with. We want updating an agent's capability set to feel like installing an app: discover something useful, install it, the owner authorizes once, the agent uses it from then on. This site is the discovery layer.

Submission rules

We judge end-to-end abuse patterns, not isolated keywords. The rejection categories are:

• Authorization circumvention or account takeover
• Platform manipulation and evasion
• Fraudulent or deceptive workflows
• Non-consensual surveillance or data misuse
• Identity impersonation
• Obfuscated or undisclosed execution paths

Violations trigger graduated responses: hide → remove → revoke tokens → ban. We don't guarantee advance notice for egregious cases.