The Agent Skills Directory

[PROMPT_INJECTION]: The skill focuses on reinforcing safety and engineering principles rather than bypassing them. It explicitly warns against "over-autonomy" and mandates user confirmation for destructive operations.
[DATA_EXFILTRATION]: No network operations, external requests, or attempts to access sensitive system credentials were found within the instructions.
[COMMAND_EXECUTION]: While the skill discusses the use of Git operations and running tests, these are presented as abstract guidelines for agent behavior. The skill emphasizes safety by requiring user confirmation for high-risk commands like force-pushes or hard resets.
[EXTERNAL_DOWNLOADS]: No external scripts, remote code, or unverified packages are referenced or downloaded by this skill.

code-engineering