Trusty Trail

Overview Trusty Trail is designed so that you can quickly check shortcuts for suspicious actions, such as deleting your data, receiving your data and sending it to servers, changing important settings, etc. The Trusty Trail script together with AI (using Nuvole Systems API, a proprietary API for Nuvole AI, Trusty Trail and other MIKL’s projects), can check whether shortcuts are safe or not by categorizing each suspicious action and generating a description for them. Direct download link Featured by Trusty Trail is featured in Blog ! In a community like, where users constantly share and download shortcuts, having tools that ensure security is essential. Trusty Trail not only protects users from potential risks but also promotes transparency and trust within the community. Additionally, its focus on user education is noteworthy. By providing clear and detailed descriptions, Trusty Trail helps users understand how shortcuts work and what actions they perform, which is especially useful for those learning to create their own shortcuts. Main Menu, Start of Shortcut Checking A convenient menu to find the shortcut you want to check. You can choose shortcut from list of all your shortcuts, using search for shortcuts, or import shortcut from link ( see the Import Shortcuts From Links section ). Also, from each menu you can quickly return to the Main page or change the search request. Import Shortcuts From Links You can check shortcuts from the Share Sheet or Main Menu from: iCloud link; link; direct download link; Post/comment from Reddit that contains above links . Additional information about importing from Reddit: If a post and the comments under it together contain multiple links, you will be able to select which shortcut you want to check. You will be shown the start of the post/comment, the post/comment author name and icon, all shortcuts with names, icons and links found in that post/comment! Result of Checking After checking, you will be shown whether the shortcut is safe or not, along with information about the shortcut and found suspicious actions. In a beautiful menu design using VCard UI and HTML with the support of both a dark theme and a light. All suspicious actions are divided into 20 categories with 3 types of danger to your device (red, orange and yellow). Here is full list of them. AI generates a description for each type found and explains what the actions found do and why shortcut is safe or not. To view these descriptions and explanations, as well as a full list of suspicious actions of a certain category, click on the check result or the desired category. More Examples of Work Error Handling and Warnings If an error occurred when requesting to the AI or the AI returned incomplete description data, this will be processed and displayed. Unfortunately, Trusty Trail has a limit on the size of the processed shortcut using AI . If the shortcut is close to the limit, there will be a warning that the AI description may not be accurate, and if the limit is exceeded, the AI check will not be performed at all. So, How Does Checking Work? Get a list of suspicious actions (this list is stored on the GitHub, so you don't need to update Trusty Trail with new OS updates). Checking for suspicious actions: Converts the shortcut to a Property List (file with a .plist extension), which contains all the shortcut actions. Checks which suspicious actions are contained in the Plist file. Create a list of suspicious actions sorted by category. Create a list of suspicious actions and categories for the AI to write a description for. Generate the JSON for AI: Retrieve a JSON file from Plist, then: Reducing action keys. Highlighting suspicious actions so that AI can notice them better. Delete large base64-encoded PNG , JPEG , GIF , BMP , WebP , SVG , PDF and M4A files. Shortcut checking with AI: Send the processed JSON to the first AI ( Gemma 4 31B provided by Google ) to check the security of the shortcut and generate a description for each category of suspicious actions. Sending a response from the first AI to the second AI ( GLM-4.7 without reasoning, provided by ultra-fast Cerebras AI API ) to correct errors and beautify the text. Compilation of the menu with result: Conversion of the second AI's response (the final result of the check) and errors handling. Sorting menu items, adding warnings and a list of suspicious actions. Badges System About badges: Using Trusty Trail, you can quickly and easily generate a badge to demonstrate your commitment to user safety and confirm your shortcut is safe to use. This badge can be easily added to your project description on or other forums. The Trusty Trail badge includes the following information: the version of Trusty Trail used to perform the check, your project's name and icon, the verification result with a brief description, a list of all suspicious actions found, and detailed shortcut details. Example of a badge: Counterfeit protection for badges: Since the badge can only be created by checking shortcut with the link, at the bottom of the badge there will be a details of which shortcut the developer checked. How to create and save a badge: Check your shortcut with Trusty Trail : You will only be able to generate a badge if you have checked your shortcut via a link or the Share Sheet and it has been verified as safe . Click on the “Generate a safety badge for shortcut” button in the menu with the result of the check. Customization: you can choose between a light or dark design theme, as well as highlighting important verification details in the color of the shortcut icon. Saving: save your badge in Markdown (for descriptions) or copy the direct link to an image, upload to Imgur, or save to Photos. Noticed an Error? If you find an error in Trusty Trail, you can easily create a bug report. You can submit a report by email or anonymously. Build-in Check for Updates You will receive notifications that an update for Trusty Trail is available. You will be shown brief changes in the update, choosing to install the update or not. By clicking on the “More Detailed Changelog” button, you will be shown the full list of changes. Compatibility OS Compatibility iOS & iPadOS 26, 18 Yes iOS & iPadOS 17 Something might not work iOS & iPadOS 16 and older No macOS / watchOS / visionOS ??? Known Issues Incorrect Trusty Trail results: Trusty Trail uses AI, which can hallucinate and give incorrect results. Be careful! Freezing at start: When you start Nuvole AI, the text entry window may not appear. Simply restart Nuvole AI. Requirement of permission to share large amouts of data: For some functions, Shortcuts app may request permission to share large amouts of data. To allow this permission, open the Settings app, navigate to Apps > Shortcuts > Advanced , then turn on the “Allow Sharing Large Amounts of Data”.