What problem does it solve? A diff that looks safe and passes CI can still break callers, data consumers, or processes it never touched. This Skill finds the downstream impact of a change and proves the safety claim it rests on by running the real code, instead of trusting comments, READMEs, or green test suites. ## Core Features & Use Cases - Impact tracing beyond grep: Follows callbacks, hooks, dynamically built keys, cross-language readers of the same data, and code several hops downstream that symbol search misses. - Executable proof of safety claims: Writes throwaway scripts that import the actual module or pinned library version and drive the path in question, so every claim is backed by a run, not an assertion. - Leveled safety reporting: Tags every claim with a confidence level from [1 asserted] to [5 live] and hands back a structured result with confirmed risks separated from checked-and-cleared items. - Use Case: Before merging a branch that renames a shared field, use this Skill to find every consumer of that field, run the real code against the pinned dependency version, and receive a report stating exactly which risks are proven and which remain unverified. ## Quick Start Ask the AI to check the blast radius of the current branch diff before merging it.