Visualizing Results
How to understand the information shown by Skyvern
Visualizing Results
Skyvern comes bundled with a Visualizer to help you understand what’s going on with your tasks. To get started, navigate to the task history page and click on any task.
Actions
Each action that Skyvern takes can be viewed in the Action viewer. It’s accompanied with a screenshot of the screen state after the action has executed
Recording
Each Skyvern task comes with a recording of the entire operation (end to end). To view the recording, click on the recording tab.
Task Parameters
The task parameters are the inputs that you provided to Skyvern. This includes the URL, the extraction rules, and any other information that you provided to Skyvern.
Diagnostic Logs
The diagnostics tab contains information used by Skyvern to do its processing. It includes information such as the annotated screenshots, action screenshots, element tree, prompt, action list, page HTML, and the raw LLM Request.
Accessing results via API
You can also access the results of a task via the API.
The following endpoint can be used to get the results of a task:
The response will contain a list of steps, alongside their actions and other relevant information
The following endpoint can be used to retrieve artifacts for a specific step:
Artifacts configurations
By default, Skyvern stores artifacts, including video recording, screenshots, llm requests and responses, html and skyvern parsed html elements locally in the /artifacts folder under the skyvern repository.
You can also have skyvern to upload atrifacts to your s3 buckets. To do this, first set up these environment variables:
AWS_DEFAULT_REGION:us-east-1,us-west-1, …AWS_ACCESS_KEY_IDAWS_SECRET_ACCESS_KEYSKYVERN_STORAGE_TYPE: set it to bes3. The default islocal
Make sure these s3 buckets are created: skyvern-artifacts, skyvern-screenshots. These are the default bucket names skyvern uses. To customize the bucket names, change these two env variables: AWS_S3_BUCKET_ARTIFACTS and AWS_S3_BUCKET_SCREENSHOTS