The Agent Skills Directory

[NO_CODE] (SAFE): The skill consists entirely of markdown instructions and templates. It does not include any scripts (Python, JS, Shell) or configuration files that execute logic or install dependencies.
[PROMPT_INJECTION] (SAFE): The use of 'CRITICAL' and 'REQUIRED' markers is strictly for output formatting and scoring logic. There are no attempts to bypass AI safety filters or override system instructions.
[DATA_EXFILTRATION] (SAFE): The skill does not perform any network operations, file reads, or credential handling. It only generates a markdown report based on provided context.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill processes untrusted code implementations as input. While a malicious implementation could attempt to manipulate its own score through embedded text, the skill lacks any capabilities (such as shell execution or network access) that could be exploited to cause harm beyond influencing the content of the evaluation report.

judge