The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions to generate stylistic alternatives when encountering sensitive or copyrighted content, aimed at preventing generation refusals. It also has an indirect prompt injection surface by including external article content in generation prompts without sanitization.
[COMMAND_EXECUTION]: Basic shell commands like test and echo are used during the initialization and workflow phases to check for the existence of the EXTEND.md configuration file and other local assets.
[SAFE]: Analysis of the skill's 31 files reveals no evidence of malicious obfuscation, data exfiltration to external domains, or attempts at privilege escalation. The skill operates as intended for automated image prompt engineering.

baoyu-cover-image