Kraken.io handles JPG, PNG, GIF, SVG, WebP, AVIF, HEIC, and PDF files. The agent can also convert between formats, for example, turning a PNG into WebP or requesting AVIF output. You specify the desired format in the conversation, and the agent passes the appropriate flags to the API.

Lossless compression reduces file size by 15-20% without any change in image quality. Lossy compression uses Kraken.io's intelligent algorithms to achieve 50-90% file size reductions while keeping visual quality indistinguishable to the human eye. The agent defaults to lossy unless you specify lossless, and you can also set a custom quality level between 0 and 100.

Yes. The Kraken.io API accepts a resize object alongside compression settings. You specify width, height, and a resize strategy (exact, portrait, landscape, auto, crop, or fill). The agent handles both operations in a single API call, returning an image that is both sized correctly and optimized for file weight.

Kraken.io returns a temporary URL for the optimized image that remains available for a limited time. You can also configure S3 storage to have Kraken.io push the optimized file directly to your Amazon S3 bucket. The agent can include s3_store parameters in the request if you want permanent cloud storage for processed images.

Sandbox mode (dev: true) processes images without consuming your API credits. The optimized output is watermarked, but file size reduction metrics are accurate. This lets you test compression settings, evaluate quality tradeoffs, and confirm resize dimensions before committing production credits. The agent can toggle sandbox mode per request.

Yes. By default, Kraken.io strips all metadata to minimize file size. If you need to preserve camera data, GPS coordinates, or copyright information, the agent sets preserve_metadata to true. This is especially useful for photographers and media companies who need to retain provenance information in their optimized images.

When the wait parameter is set to true, Kraken.io processes the image synchronously and returns results in the same response. Typical processing time ranges from 1 to 5 seconds depending on file size and optimization mode. For larger images or batch operations, the agent can process asynchronously and use a callback webhook.

Kraken.io's web interface requires manual uploads and downloads for each image. With Tars, you paste a URL or describe what you need, and the agent handles API calls, settings, and delivery. It integrates into your team's Slack, Teams, or website workflow, making optimization a natural part of the content pipeline instead of a separate step.