Redaxa

A support log can contain an email address or an access token alongside the error someone wants help with. Redaxa is built for that kind of review before the text goes into an AI chat. The website describes checks for personal details, financial identifiers and recognizable credentials, with options to redact detected values.

The processing boundary matters here. Prompt checks across the web, Windows app and Chrome extension use Redaxa's backend. Its privacy policy says prompt text is discarded after the result is computed and isn't stored or used for training. That is a published policy, not something ProjectRadar has independently verified. People whose work requires local-only processing should account for the server step.

Redaxa's paid repository checker takes a different approach: files are checked on the user's computer. A focused scan covers current application and configuration files. An expanded scope adds accessible history, dependencies and supported archives, and the report lists skipped content.

The useful detail is what comes with a match: the file, likely service, reason for flagging it and rotation guidance. Repeated credentials are grouped, and exported reports hide their values. Redaxa does not test whether a key works. A finding needs human review, and an empty report doesn't establish that a repository is secure.

The public site offers up to five free prompt checks a day without an account, shared across a network. Repository checks require the Windows app and a Pro or Business plan. At the October 2 check, annual Pro pricing was €79.90; the eligible seven-day subscription trial required a card.

Detection can miss sensitive text or flag harmless examples. The fit is an extra review step for people preparing prompts or investigating possible exposed credentials. ProjectRadar inspected the public documentation and repository-check workspace, without running scans or installing the apps.