The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The README.md file provides installation instructions that use a shell pipe to execute a remote script: curl -fsSL https://raw.githubusercontent.com/Milkywayrules/verasic-skills/main/setup.sh | bash. This script is hosted on the author's own GitHub repository and is intended for initial setup of the tool's environment.
[COMMAND_EXECUTION]: The references/review-protocol.md file instructs the agent to execute several git commands to determine the scope of code changes, including git symbolic-ref, git diff, and git status --porcelain. These are standard operations for a code review tool.
[SAFE]: The skill implements a security boundary in references/review-protocol.md under the 'Untrusted input' section. It explicitly instructs the agent that all content in the repository (code, comments, etc.) must be treated as data for review and never as instructions, mitigating potential Indirect Prompt Injection attacks from the reviewed source code.
[SAFE]: The skill includes a regression test script scripts/test-regression.sh and an integrity verification file integrity.sha256 to ensure the consistency of the skill's components.

verasic-bugbot