Typescript Sdk

The TypeScript SDK skill enables AI agents to interact with the Comet ML Opik platform, allowing them to log experiments, track metrics, and manage model evaluations directly from code. It provides a programmatic interface for developers to integrate experiment tracking into their machine learning workflows without manual setup.

When to use it

• Automating the logging of training runs and hyperparameters during CI/CD pipelines.
• Retrieving historical experiment data to compare model performance across different versions.
• Triggering automated evaluations on new datasets once a model reaches a specific accuracy threshold.
• Generating structured reports of model metrics for stakeholder review without exporting raw logs.

Key capabilities

• Logging experiments with custom metadata and tags.
• Tracking scalar metrics, images, and plots over time.
• Managing projects and datasets within the Opik workspace.
• Querying experiment results via SDK methods.

Example prompts

• "Log this training run with hyperparameters learning_rate=0.01 and batch_size=32 to my current project."
• "Retrieve all experiments from last week where the loss was below 0.5 and show me their final metrics."
• "Create a new dataset entry for 'customer_churn_v2' and upload the CSV file located at /data/churn.csv."

Tips & gotchas

Ensure your environment has the opik package installed before attempting to use this skill, as it relies on specific authentication tokens. The SDK requires valid API credentials configured in your local environment or passed securely via context variables to access private projects.