Log Type Reference

Log Types

All logs have the following top-level, shared metadata:

Timestamp: 2024-12-13 18:42:15.924

Log level: Debug

Log type: action.http

Plugin name: Lookup Customer Account Details

Requestor: [email protected]

Summary: HTTP request executed

Data redaction: Sensitive data in this log is not redacted. Learn more

plugin.filter

Example:

"root":{
  "created_at":"2025-04-14T03:08:12Z"
  "plugins_passed_filter":{...}
  "rewritten_utterance":"<REDACTED>"
  "user_utterance":"<REDACTED>"
}

🚧

Attributes labeled with a ❗️are always redacted in this log

Attribute Descriptions:

  • plugins_passed_filter: this will be a list of the Plugins that passed the bots filtering process.
  • rewritten_utterance: the rewritten user utterance that is sent to the LLM.
  • user_utterance: the original user utterance from chat.

What issues does this help debug?

  • If your Plugin is low in the list but not selected, you may need to write better names, descriptions, and example triggering utterances
  • If your Plugin is not on the list at all, you may not be passing the Launch Permissions that are set on the Plugin.

plugin.trigger

Example:

"root":{
  "created_at":"2025-04-14T03:08:12Z"
  "error_message":""
  "selected_plugin":"KnowledgeSearchPlugin"
  "slots":{...}
}

🚧

Attributes labeled with a ❗️means they can be redacted. Learn more about log redaction here.

Attribute Descriptions:

  • error_message: Optional error message if there was a failure in triggering your Plugin.
  • selected_plugin: This is the plugin that was selected to run
  • slots: A list of slot names & the associated values that were resolved from the user.

What issues does this help debug?

  • Why did my Plugin fail to trigger?
  • Are the slot values correct?

plugin.response

Example:

"root":{
  "bot_input":""
  "created_at":"2025-04-14T03:08:18Z"
  "error_message":""
  "execution_id":""
  "message":"<REDACTED>"
  "message_type":"MESSAGE_TYPE_FINAL"
  "plugin_id":"ce815b89-f716-41dd-b998-63bb5d606b1e"
  "plugin_status":"PLUGIN_STATUS_DONE"
}

Attribute Descriptions:

  • error_message: Optional error message if there was a failure in sending a response from your Plugin.
  • message: the message that was sent from the bot to the user.

What issues does this help debug?

  • Why did my bot not send a message back to the user?

program_plugin.execution

Example:

"root":{
  "created_at":"2025-04-14T03:08:15Z"
  "execution_id":"6aFu3ILfBbGB"
  "execution_updates":[...]
  "plugin_id":"ce815b89-f716-41dd-b998-63bb5d606b1e"
  "plugin_name":""
  "plugin_status":"PLUGIN_STATUS_PROCESSING"
}

Attribute Descriptions:

  • execution_updates: All of the data related to the run of an Activity.

What issues does this help debug?

  • Why is my Plugin not working?

compound_action.trigger

Example:

"root":{
  "accessed_variables":{
    "request_type":"get"
    "test_user":{...}
  }
  "created_at":"2024-12-13T18:03:28Z"
  "error_message":""
  "input_arguments":{
    "request_type":{...}
    "test_user":{...}
  }
  "progress_updates":[
  ]
  "return_value":"null"
  "root_uuid":"4vzrjqXHKCMd"
  "status":"WORKFLOW_STATUS_PENDING"
  "step_statuses":{
    "mfm_workday_time_off_data_retrieval":"WORKFLOW_STATUS_PENDING"
    "mfm_workday_time_off_data_retrieval.1-be16ed24-2b86-401c-be91-867dbfe9dd9b":"WORKFLOW_STATUS_PENDING"
  }
  "use_case_uuid":"51c834b4-d98c-4f3a-9739-15a9c87c7322"
  "workflow_id":"vuX-0PRliibN"
}

Attribute Descriptions:

🚧

Attributes labeled with a ❗️means they can be redacted. Learn more about log redaction here.

  • ❗️accessed_variables - The variables that were accessed during the triggering of the compound action
  • created_at - The time when the log was emitted.
  • error_message - A Moveworks error message. Only applicable if an error was thrown.
  • ❗️input_arguments - This a dictionary of input arguments used to trigger the compound action. Each item in the dictionary details the input argument name, data type, and value.
  • progress_updates - A list of Progress Updates that were sent back to the end user. This will be empty.
  • ❗️return_value - This will always be null
  • root_uuid - The root uuid that ties all of the compound_action.trigger, compound_action.step,action.http together for a single execution of a Plugin.
  • step_statuses - This is a dictionary of all of the steps & their respective statuses.
  • status - This is the first status of the compound action. Can be PENDING or ERROR.
  • use_case_uuid - This is the UUID of the compound action we are triggering.
  • workflow_id - This is the compound action's instance execution ID. It will be unique for each run of the compound action.

What issues does this help debug?

  • Compound Action was triggered with incorrect/malformed data
  • Compound Action threw a Moveworks error

compound_action.step

"root":{
  "accessed_variables":{
    "request_type":"get"
    "mfm_workday_get_timeOffDetails_result":{...}
    "mfm_workday_get_worker_result":{...}
    "mfm_workday_get_eligibleAbsenceTypes_result":{...}
    "test_user":{...}
  }
  "created_at":"2024-12-13T18:03:48Z"
  "error_message":""
  "from_status":"WORKFLOW_STATUS_PENDING"
  "payload_data":{
    "progress_update":{}
    "error":{}
    "result":{...}
    "status":"ACTION_STATUS_DONE"
    "uuid":"1FSDh_S7z6kF"
    "result_schema":{...}
  }
  "progress_updates":[
  ]
  "return_value":{
    "allTimeOffs":[...]
    "timeOffTypes":[...]
    "instructions_for_model":"You MUST only share what the user asked about"
  }
  "root_uuid":"4vzrjqXHKCMd"
  "step_statuses":{
    "mfm_workday_time_off_data_retrieval.3":"WORKFLOW_STATUS_COMPLETE"
    "mfm_workday_time_off_data_retrieval.3.2.2":"WORKFLOW_STATUS_COMPLETE"
  }
  "to_status":"WORKFLOW_STATUS_COMPLETE"
  "workflow_id":"vuX-0PRliibN"
}

Attribute Descriptions:

🚧

Attributes labeled with a ❗️means they can be redacted. Learn more about log redaction here.

  • ❗️accessed_variables - The variables that were accessed during this step of the compound action.
  • created_at - The time when the log was emitted.
  • error_message - A Moveworks error message. Only applicable if an error was thrown.
  • from_status - This is the previous status of the compound action
  • ❗️payload_data - All of the data that was sent and received for an Action. Only applicable if an Action was triggered at this step.
  • progress_updates - A list of Progress Updates that were sent back to the end user.
  • ❗️return_value - The final data object that was sent to the LLM.
  • root_uuid - The root uuid that ties all of the compound_action.trigger, compound_action.step,action.http together for a single execution of a Plugin.
  • step_statuses - This is a dictionary of all of the steps & their respective statuses.
  • to_status - This is the new status of the compound action. Can be
  • workflow_id - This is the compound action's instance execution ID. It will be unique for each run of the compound action.

What issues does this help debug?

  • Compound Action hit the wrong conditional
  • HTTP, Script, or Built-in Action failed
  • Compound Action isn't returning the correct data
  • Input/Outputs of each step are incorrect/malformed

action.http

Example:

"root":{
  "canonical_endpoint":"moveworks.my.salesforce.com/services/data/v58.0/query?q=S*****'"
  "request_headers":"Authorization: B*****h"
  "request_method":"GET"
  "request_payload":"<EMPTY>"
  "response_code":{}
  "response_payload":{...}
}

Attribute Descriptions:

🚧

Attributes labeled with a ❗️are always redacted in this log type.

  • canonical_endpoint - The full Base URL + Endpoint we made a request to.
  • ❗️request_headers - The headers sent with the request.
  • request_method - The request method. This can be GET, POST, PUT, DELETE, or PATCH.
  • ❗️request_payload - The request body sent with the request.
  • response_code - The response code returned from the request (e.g. 200, 404).
  • ❗️response_payload - The response returned from the request

What issues does this help debug?

  • HTTP Request failed due to bad/incorrect auth info
  • HTTP Request failed due to bad/incorrect request info