QuoDroid Software Development Private

Ask how your ServiceNow instance works, then update config with a plan-and-approve workflow. Powered by Claude.

TorchTiger Engineer answers questions about the business rules, scripts, flows and ACLs already on your instance, and traces how they connect. Write mode is off by default. To change something, it proposes a plan, checks every field against your instance, and creates, updates or deletes nothing until you approve. After it writes, it reads each record back, and you can undo updates and new records from the change log. On production it switches off completely. It runs on Claude with your own Anthropic API key. It's free, with no TorchTiger account; you pay Anthropic for your usage. Your key stays encrypted on your machine under a passphrase only you know, and requests go from your browser straight to Anthropic: no TorchTiger server is involved, and we never see your key or your data. It uses your own signed-in session and permissions: no password to enter, no platform plugin. WHAT IT DOES 1. Ask anything about your instance Ask in plain English. It searches business rules, client scripts, script includes, UI policies and flows, then traces how they connect. It knows which record you're on. Ask "where is this used?" and it looks in 27 places at once, with a count for each. 2. Approve every change first Turn Write mode on and the AI proposes a plan first: nothing it plans is created, updated or deleted until you approve. Before you see the plan, it checks every field and marks any that may not save; after it writes, it reads each record back. A step only the record's form can do runs in your own tab while you watch. On production there's no chat, no reads and no writes, and it stays off on any instance it can't confirm is non-production. 3. See every change in the log The change log records every change the AI makes, grouped by update set, so you know what was created, updated or deleted. Update sets are named your way: the date, your initials, the story number and a short description. 4. Undo from the log Select a logged change and undo it: Engineer reverses it from what it saved at the time, after you approve. Updates and new records revert. A delete, a change to who can do what, data outside update sets or a change to many records at once can't be reverted, so it asks for your confirmation again when it's reached. 5. Check the result on the page After a change people see (a catalog item, a form, a portal page), it can open the page and take a picture to check the change shows. Pictures go to your AI provider with the conversation, can't be redacted, and aren't kept. By default Chrome asks for one click on the toolbar icon per tab; the Page checks switch in the popup asks once for the optional permission instead. 6. Work app by app The panel shows the app you're working in: the whole instance, until you pick an app or take the one the page you're on belongs to. Answers, plans and changes stay in that app. When a change can't be made here, you get a write-up for the person who can make it. 7. Commit and publish For an app you develop, review the changes that aren't in Git yet, then commit them, publish a version to your company's app repository, or both. With admin rights and a GitHub repository, it also saves an installable update set file of each version in Git. Nothing runs until you click. 8. Check Store readiness Before you submit an app to the ServiceNow Store, run 15 rule checks drawn from the Store certification self-test and the platform's scan checks. You get a score, the checks that block submission, and a fix for each finding. The checks run in your browser, on your app's records. Where the Certification Self-Test Tool is installed, run it from the panel and decide each failure. It doesn't replace the Store's own certification review. If you ask for paid help, the chat can link to XpertApps, paid app-development help from TorchTiger's maker, and to ServiceNow's Partner Finder. WHAT YOU NEED - Google Chrome or Microsoft Edge - A ServiceNow instance you can sign in to (a personal developer instance works) - An API key from the Claude Console, pay-as-you-go. Claude Pro/Max subscription tokens aren't supported: Anthropic doesn't allow third-party tools to use them. GET STARTED Install it and connect your own API key; the chat panel opens on any instance page. Set-up guide: https://torchtiger.com/get-started.html Demo videos: https://torchtiger.com/videos/ TorchTiger Engineer is independent of ServiceNow, Inc. and of Anthropic, PBC. ServiceNow is a trademark of ServiceNow, Inc., and Claude is a trademark of Anthropic, PBC.