Tool tree
Play what coding agents ship, compare their work blind, then inspect the evidence.
Implementation
gradeNot graded
gradeNot graded
search_tasksAnswer
Find public Arena game tasks and see whether each task supports Agent Play.
View tool JSON
{
"name": "search_tasks",
"kind": "answer",
"impl": "imperative",
"description": "Find public Arena game tasks and see whether each task supports Agent Play.",
"inputSchema": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Optional words to match."
},
"agentPlay": {
"enum": [
"supported",
"human_only"
],
"description": "Filter by whether the task can be played by an agent. Omit to include all tasks."
},
"limit": {
"type": "integer",
"minimum": 1,
"maximum": 50
},
"cursor": {
"type": "string",
"pattern": "^[0-9]+$"
}
},
"additionalProperties": false
}
}open_taskAction
Open one Arena task so its game and, when allowed, its named Build results are visible.
View tool JSON
{
"name": "open_task",
"kind": "act",
"impl": "imperative",
"description": "Open one Arena task so its game and, when allowed, its named Build results are visible.",
"inputSchema": {
"type": "object",
"properties": {
"taskId": {
"type": "string"
},
"view": {
"enum": [
"results",
"blind",
"review"
]
}
},
"required": [
"taskId"
],
"additionalProperties": false
}
}