Troubleshoot skill trainer issues

Available in Classic and VPC

The following describes issues that may occur while using the skill trainer and how to resolve them.

Troubleshoot skill set response format issue

The skill set response format has been configured, but it is not applied correctly to the final responses during data collection.

Cause

Depending on the complexity of the skill and user query, the configured response format may not be applied correctly to the final response.

Solution

Try performing training after directly editing the final response on the data collection screen.
For example, if you set the response format of the skill set to "Format: JSON", but the actual final response generated by the model is in the form of "Final response A", edit it to "Final response B" and save it. To yield improved responses, try collecting multiple instances of this type of data and then start training.

• Final response A

  As of August 9, 2025, the number of purchase conversion users is as follows: 1,003,120 page views and 1,222,334 users.
  

• Final response B

  {
     "20250809":{
        "purchase":[
           {
              "pv":"1,003,120",
              "user":"1,222,334"
           }
        ]
     }
  }
  

The action input does not produce the intended result

The action input does not produce the intended result.

Cause

If the API spec is not written in detail, results that do not match the intended purpose may be generated.

Solution

Add an example to the 'description' of the parameters in API Spec. With examples, the language model can understand user queries based on API Spec and generate URLs with correct parameters.
Example of writing:

{
    "keyword": {
        "type": "string",
        "description": "A modifier in the form of an adjective or an adverb. Example: Kind, satisfactory, good",
    },
    "review": {
        "type": "string",
        "description": "Positive or negative assessment. Example: Either pos or neg"
    }
}

Error message: "No plannable skills"

During data collection, the error message "No plannable skills" is displayed.

Cause

This may occur if the skill has not been saved.
This may occur if the model does not correctly understand the skill's API spec or manifest.

Solution

• Check whether the skill has been saved in the 'completed' status.
• Please rewrite them according to the API Spec and Manifest writing guide to ensure that the model can understand them sufficiently. If the problem persists, please leave an inquiry on the Forum.

Error message: "Token limit exceeded"

During data collection, the error message "Token limit exceeded" appears.

Cause

During data collection, the maximum number of tokens per step is limited to 8,192.

Solution

For the sum of the tokens per step, refer to the following information. To calculate the number of tokens, you can use Explorer > Token calculator (HCX) in CLOVA Studio.

• Query analysis: System prompt, user query, Manifest, query analysis result
• Skill calls
  • Step 2 (viewing API search result): System prompt, call method, call skill, API query details, API search result, action, action input
  • Step 3 (call result): System prompt, call method, call skill, API view details, API search result, action, action input, call result, thought, and result
• Final response: System prompt, result, final response

Reset call options

Call options are reset each time data is collected.

Cause

After collecting data with the applied call options, the call options remain unchanged even if the [Reset] button is clicked for the next data collection. However, if you exit the data collection screen and return later, the call options will be reset.

Solution

If the call options are reset when returning to the data collection screen, use the Load function to retrieve the temporarily saved task. Previously used call options can also be loaded together.

"Data collection log" error

An error has occurred during data collection task.

Solution

If an error occurs during a task, contact Customer Support with the following information.

• Screenshot of the screen where the error occurred
• Skill set
• Skill information

Data collection was executed, but the task did not proceed

Even after writing both the API Spec and manifest, and entering the user query, the task does not proceed when executed.

Cause

• There is an issue with the API Spec.
• The method of providing parameters in the API spec is incorrect.

Solution

See the Creating API Spec and Skill trainer use cases to review the API Spec and its parameters.

API response delay when using the skill trainer

The API response takes too long when using the skill trainer.

Cause

The skill trainer is affected by the processing speed of the integrated API. Regardless of the skill engine speed, response time may be slow if the actual API being called is complex or if multiple skills (APIs) need to be invoked at once depending on the user request.

Solution

• Check whether the actual API being called is overly complex.
• Modify it to prevent multiple skills (API) from being called at once.
• Identify the section where the delay occurs and provide it to Customer Support.