Soracom Flux

Collect Messages Action

The Collect Messages action aggregates messages from branched processing flows in your Flux app. When actions are branched using the Output > Advanced Settings > Send event for each option, this action collects the results back into a single message.

For example, you can execute RTSP Camera image action and AI actions targeting multiple cameras, then use the Collect Messages action to aggregate the results before sending a single consolidated notification.

Configuration

Config

Configure the Collect Messages action settings:

Collect Target Channel

Aggregates messages sent to the specified message aggregation target channel.

Destination Channel

Specify the destination channel for aggregated messages. Once all messages from the branched processing have been received, they are aggregated and sent to the destination channel.

Advanced Settings: Anchor

When a Flux app has multiple branches, you specify the channel or action that determines the starting point for aggregation. The resource serving as this starting point is called an anchor. This action executes once all processing after the anchor has completed.

In a Flux app, you can configure one or more actions on a channel or send the execution result (output data) of an action to another channel. Therefore, the paths included in the aggregation depend on which channel/action is set as the anchor.

• Anchor Behavior Example
  • You can select either an action or a channel as an anchor. The behavior changes depending on the anchor type and placement, as follows:
  • When an action is set as the anchor
  • Only channels/actions following that action become aggregation targets if the action itself executes.
  • Even if this action or subsequent processing fails and messages do not reach the aggregation target channel, the aggregation process itself may still run, potentially returning an empty array.
  • When setting a channel as the anchor
  • Only channels/actions following that channel become aggregation targets if a message reaches that channel.
  • If the message does not reach this channel due to condition mismatch or error, the aggregation process will not start.
  • Guidelines for Usage
  • When you want to aggregate "only the cases where a specific action was executed": Set that action, or the channel/action following it, as the anchor.
  • When you want to aggregate "always," including cases where the action is skipped due to condition mismatch: Set the channel/action preceding that action as the anchor.

Output

Configure how to handle the output data of the action. Refer to Enable Republishing of Action Output for more details.

Output Data of the Action

The Collect Messages action outputs an array containing all collected messages in the messages field.

Example: Collecting Soracom Cloud Camera Images

The following example shows the output when multiple Soracom Cloud Camera image acquisition actions are executed and collected:

{
  "messages": [
    {
      "deviceId": "7CDDE90XXXXX",
      "imageUrl": "https://flux-core-temporary-object-storage-jp-prod.s3.ap-northeast-1.amazonaws.com/01KCTRENXXXXXXX",
      "exportId": "5d06a3a4-XXXX",
      "harvestFilesPath": "/flux/XXXX.jpg",
      "datetime": "2025-12-19T07:28:35.035Z"
    },
    {
      "deviceId": "7CDDE90YYYYY",
      "imageUrl": "https://flux-core-temporary-object-storage-jp-prod.s3.ap-northeast-1.amazonaws.com/01KCTRENYYYYYYY",
      "exportId": "5d06a3a4-YYYY",
      "harvestFilesPath": "/flux/YYYY.jpg",
      "datetime": "2025-12-19T07:28:36.017Z"
    }
  ]
}

Example: Collecting AI Action Results

The following example shows the output when multiple AI actions are executed and collected:

{
  "messages": [
    {
      "output": {
        "text": "<AI response>"
      },
      "usage": {
        "completion_tokens": 246,
        "prompt_tokens": 3949,
        "total_tokens": 4195,
        "prompt_tokens_details": {
          "cached_tokens": 0,
          "audio_tokens": 0
        },
        "completion_tokens_details": {
          "reasoning_tokens": 0,
          "audio_tokens": 0,
          "accepted_prediction_tokens": 0,
          "rejected_prediction_tokens": 0
        },
        "model": "gpt-4.1-nano",
        "byol": false,
        "service": "openai",
        "credit": 2
      }
    },
    {
      "output": {
        "text": "<AI response>"
      },
      "usage": {
        "completion_tokens": 244,
        "prompt_tokens": 3949,
        "total_tokens": 4193,
        "prompt_tokens_details": {
          "cached_tokens": 0,
          "audio_tokens": 0
        },
        "completion_tokens_details": {
          "reasoning_tokens": 0,
          "audio_tokens": 0,
          "accepted_prediction_tokens": 0,
          "rejected_prediction_tokens": 0
        },
        "model": "gpt-4.1-mini",
        "byol": false,
        "service": "openai",
        "credit": 6
      }
    }
  ]
}

The attributes are as follows: