The Agent Skills Directory

Dynamic Execution of Diagnostic Probes: The skill is designed to generate and execute temporary TypeScript scripts to verify how code behaves at runtime. This capability is managed by a strict requirement for explicit user approval of the exact code and command before every execution, and instructions to use temporary directories for isolation.
Access to Sensitive Environment Variables: To test live integrations, the skill can read environment variables such as OPENAI_API_KEY. It includes robust protections against data exposure, such as a prohibition on printing secret values and a mandatory disclosure step where the user must approve the specific variables being read.
Command Line Tool Usage: The skill utilizes local development tools including git, pnpm, tsc (TypeScript compiler), and tsx (TypeScript executor). These are used to establish runtime context and run probes within the repository's environment, focusing on observed evidence rather than persistent changes.
Structured Safety Process: A comprehensive workflow is enforced that requires planning, disclosure of capabilities (network, filesystem, host services), and explicit approval gates for mutating or costly operations, effectively placing the agent under direct user supervision for all high-risk tasks.

runtime-behavior-probe