Paste your test titles (and tags, if you track them) and get near-duplicate clusters flagged by token-overlap similarity — a real, deterministic algorithm (Jaccard similarity, the standard "how similar are these two sets" measure), not an ML guess. Worst pairs first, so a suite audit starts with the clearest candidates. Free.
One per line: title or title | tag1, tag2. Blank lines and # comments are ignored.
Unlimited clustering is free and stays free. The Slimming Playbook adds what a cluster list can't: how to actually merge or retire what it finds, safely.
Team licence for an engineering org? Email for a group rate.
Unlocks the Slimming Playbook on this device and survives reloads. No account, nothing transmitted.