The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts compare.py, generate_variant.py, and optimize_loop.py utilize the subprocess.run function to orchestrate local development tools such as the Go test runner, Git for worktree isolation, and the Claude CLI for automated variant generation and behavioral testing. These operations are conducted without shell execution, which prevents command injection and aligns with the skill's stated purpose of providing an automated benchmarking environment.
[SAFE]: Analysis of the instructions and supporting scripts confirms the absence of malicious patterns. There are no attempts at data exfiltration, credential harvesting, or remote code execution. The skill follows established development practices and utilizes local project data for its optimization tasks.

agent-comparison