The Agent Skills Directory

Command Execution Patterns: The skill instructions suggest using Bash to execute code with adversarial inputs (e.g., malformed strings, boundary cases). This is a functional requirement for an adversarial testing tool, used to confirm whether hypothesized vulnerabilities or crashes actually occur in the code being analyzed.
Input Processing Surface: The skill is designed to ingest and analyze untrusted code provided by users to find flaws. While this is necessary for its purpose, it is worth noting that the agent will be processing and potentially executing test cases against this external data as part of its 'breaker' workflow.

tester-breaker