The Agent Skills Directory

[SAFE]: The skill implements a documented prompt engineering technique aimed at improving model performance on specific tasks like long-context retrieval and multiple-choice questions.
[NO_CODE]: Although the skill includes a Python script (prompt_repetition_transformer.py), the code is entirely self-contained logic for string manipulation and does not engage in network activity, sensitive file access, or remote code execution.
[COMMAND_EXECUTION]: The requested tools (Bash, Read, Write, Grep, Glob) are appropriate for a skill focused on analyzing and transforming prompt files. No malicious command injection or privilege escalation patterns were found.
[DATA_EXFILTRATION]: The skill does not contain any hardcoded credentials, sensitive file paths, or network requests to non-whitelisted domains.

prompt-repetition