The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/run-evals.py uses subprocess.run to invoke the claude CLI. This is a functional requirement of the skill's regression harness to evaluate the agent's performance in a sandbox environment.
[EXTERNAL_DOWNLOADS]: The skill references documentation and source code from twoslash.netlify.app and github.com/twoslashes/twoslash. These are well-known official sources for the technology the skill describes.
[COMMAND_EXECUTION]: The skill documentation includes instructions for users to run local scripts using uv run. These scripts (inspect-markers.py and run-evals.py) were analyzed and found to perform their stated functions (marker alignment checking and eval orchestration) without malicious side effects.

twoslash