If you've ever stared at a feature with five parameters and realized there are hundreds of possible combinations to test, this is the methodical way out. It takes your requirements or code, identifies test parameters and their values, then generates a PICT model with constraints to produce pairwise test cases. You get the model file, a markdown table of actual test scenarios, and expected outputs for each. The real value is in how it thinks through the analysis: equivalence partitioning, boundary values, which combinations are invalid and why. It typically cuts your test suite by 80-90% compared to exhaustive testing while still catching interaction bugs. Works well for login flows, API endpoints, config matrices, anywhere combinatorial explosion is your enemy.
npx skills add https://github.com/omkamal/pypict-claude-skill --skill pict-test-designer