Tool tree

motion-arena.vercel.app

AI & AgentsTools verifiedOct 26, 2026

A browser-native motion reconstruction arena for the OpenAI WebMCP Challenge.

inspect_episodeAction

Inspect the instruction, hidden interval, execution mode, scores, budget, and allowed search ranges.

View tool JSON

{
  "name": "inspect_episode",
  "kind": "act",
  "impl": "imperative",
  "description": "Inspect the instruction, hidden interval, execution mode, scores, budget, and allowed search ranges."
}

run_rolloutAction

Generate and verify one candidate with MoMask search parameters; the LEFT motion updates visibly.

View tool JSON

{
  "name": "run_rollout",
  "kind": "act",
  "impl": "imperative",
  "description": "Generate and verify one candidate with MoMask search parameters; the LEFT motion updates visibly.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "seed": {
        "type": "number",
        "minimum": 0,
        "maximum": 2147483647
      },
      "temperature": {
        "type": "number",
        "minimum": 0.1,
        "maximum": 1.5
      },
      "condScale": {
        "type": "number",
        "minimum": 0.5,
        "maximum": 8
      },
      "topK": {
        "type": "number",
        "minimum": 1,
        "maximum": 100
      },
      "timeSteps": {
        "type": "number",
        "minimum": 4,
        "maximum": 64
      },
      "residualTemperature": {
        "type": "number",
        "minimum": 0,
        "maximum": 1.5
      },
      "residualCondScale": {
        "type": "number",
        "minimum": 0.5,
        "maximum": 8
      }
    }
  }
}

inspect_rewardAnswer

Read deterministic aggregate verifier signals. Ground-truth arrays and hidden tokens are never returned.

View tool JSON

{
  "name": "inspect_reward",
  "kind": "answer",
  "impl": "imperative",
  "description": "Read deterministic aggregate verifier signals. Ground-truth arrays and hidden tokens are never returned.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "rolloutId": {
        "type": "string"
      }
    }
  }
}

refine_spanAction

Generate and verify a candidate targeted at one normalized temporal span.

View tool JSON

{
  "name": "refine_span",
  "kind": "act",
  "impl": "imperative",
  "description": "Generate and verify a candidate targeted at one normalized temporal span.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "start": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "end": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "seed": {
        "type": "number",
        "minimum": 0,
        "maximum": 2147483647
      },
      "temperature": {
        "type": "number",
        "minimum": 0.1,
        "maximum": 1.5
      },
      "condScale": {
        "type": "number",
        "minimum": 0.5,
        "maximum": 8
      },
      "topK": {
        "type": "number",
        "minimum": 1,
        "maximum": 100
      },
      "timeSteps": {
        "type": "number",
        "minimum": 4,
        "maximum": 64
      },
      "residualTemperature": {
        "type": "number",
        "minimum": 0,
        "maximum": 1.5
      },
      "residualCondScale": {
        "type": "number",
        "minimum": 0.5,
        "maximum": 8
      }
    },
    "required": [
      "start",
      "end"
    ]
  }
}

submit_bestSensitive Action

End the episode and submit the best aggregate-scoring rollout.

View tool JSON

{
  "name": "submit_best",
  "kind": "transact",
  "impl": "imperative",
  "description": "End the episode and submit the best aggregate-scoring rollout.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "rolloutId": {
        "type": "string"
      }
    }
  }
}