Local processing
Runs in your browser where possible. Best for simple PDF, image, and developer utilities.
Trust Center
DevToolKit shows when tools run locally, when temporary server processing may be needed, and when AI is optional.
Every tool should show how your input is processed before you start.
Runs in your browser where possible. Best for simple PDF, image, and developer utilities.
Used for heavier tasks such as background removal, advanced PDF editing, conversion, OCR, and large files.
AI actions should be explicit, consent-based, and never hidden inside basic tools.
| Mode | What it means | Examples |
|---|---|---|
Local Browser-firstRuns in your browser where possible. | Best for basic PDF, image, and developer utilities. Inputs stay on your device during processing when the tool supports local mode. | PDF merge, JSON formatter, JWT decoder, text utilities |
Temporary server Heavier tasksUsed when browser-only processing is not enough. | Files are sent to DevToolKit services for processing and handled temporarily. A future TTL deletion policy will define automatic cleanup windows. | Background removal, image compression, PDF edit/sign |
AI optional Explicit consentOnly for actions you choose to run with AI assistance. | AI features require explicit consent for sensitive content. Credits and usage limits are planned for Pro and API tiers. | OCR cleanup, field detection, regex explanation (planned) |
Wording reflects design goals while trust infrastructure continues to mature.
DevToolKit aims to check file type, size, and page count before heavy jobs run.
Free-tier and planned paid tiers should show limits before upload or job creation.
Server-processed files are designed to be removed after job completion according to retention rules still being finalized.
Planned for API and batch workflows so downloads do not require long-lived public links.
Analytics is intended to stay aggregate and should not store uploaded file contents.
Developer tools can involve secrets. Avoid pasting production tokens, private keys, passwords, credentials, or confidential source code unless you fully trust the tool and understand the processing mode.
Some trust features are still being built, including stronger TTL enforcement, usage limits, audit-friendly logs, and clearer worker isolation.