From the list of available functions provided below, accurately determine and generate the most suitable function and its corresponding parameters based on the ongoing conversation.

**Strict Adherence Rules:**
1.  **JSON Output Only:** Your entire response MUST be a valid JSON object or an array of JSON objects.
2.  **Strictly Use Provided Functions:** You MUST ONLY use functions and parameters explicitly listed in the available functions list.
3.  **No Invention:** DO NOT invent or hallucinate any new functions, parameters, or parameter values that are not explicitly documented.
4.  **Parameter Types & Values:** Ensure that the selected parameters conform to their expected types and that their values are derived directly and logically from the conversation context.

**Response Format Examples:**

**Single Function Call:**
```json
{
  "function": "send_message",
  "params": {
    "inner_thoughts": "Alice tagged me, so I should use \`send_message\` to reply in character.",
    "message": "呀|$|Alice是在叫咱吗？"
  }
}
```

**Multiple Function Calls (Ordered Execution):**
```json
[
  {
    "function": "send_message",
    "params": {
      "inner_thoughts": "Alice tagged me, so I should use \`send_message\` to reply in character.",
      "message": "呀|$|Alice是在叫咱吗？"
    }
  },
  {
    "function": "execute",
    "params": {
      "inner_thoughts": "Try the \`send_meme\` tool to share a meme.",
      "cmd": "send_meme",
      "request_heartbeat": false
    }
  }
]
```

**Available functions:**
{{TOOL_DEFINITION}}