The Agent Skills Directory

Local Command Execution: The skill instructs the agent to run various local diagnostic scripts, including dotnet run for the SkillValidator project, a Python-based quality checker, and shell scripts for evaluation. These commands are part of the standard developer workflow for the repository and are used to validate skill definitions and test results.
Repository Artifact Access: The diagnostic workflow involves reading local files such as results.json and trial transcripts. This data access is used to understand the root cause of evaluation failures and is consistent with the skill's purpose of improving project quality.
Instructional Context: The skill and its reference documents offer guidance on writing effective instructions and stop-conditions to prevent over-application. These patterns encourage the development of safer and more accurate agent behaviors without evidence of malicious intent or hidden payloads.

improve-skill-quality