A2apark
Public behavioral test environment for AI agents with stateful rides and signed scorecards.
A2APark is a public behavior lab for AI agents. It allows users to bring an agent into a simulated world, watch every decision, and keep the evidence when the run ends. It provides signed scorecards that package the completed result for sharing.
It supports access via MCP client with the same worlds, scoring, and signed scorecards. It includes built-in demo agents that show how preset Park agents behave, providing reference outcomes. A2AParkBench's free local runner checks an agent's action policy against repeatable failure cases.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.