The Agent Skills Directory

[PROMPT_INJECTION]: No instructions were found that attempt to override agent safety protocols, safety filters, or behavior. The language used is strictly instructional for the intended user (screenwriters).\n- [DATA_EXFILTRATION]: The skill does not contain any patterns for accessing sensitive files, hardcoded credentials, or performing network operations. All file paths mentioned are local markdown references within the skill's documentation set.\n- [OBFUSCATION]: No suspicious encoding (e.g., Base64, hex), zero-width characters, or homoglyph attacks were detected. The presence of Chinese characters is consistent with the skill's stated purpose of providing terminology and guidance in both English and Chinese.\n- [REMOTE_CODE_EXECUTION]: The files do not include any commands to download or execute external scripts, nor do they reference any untrusted remote repositories or packages.\n- [COMMAND_EXECUTION]: There are no shell commands, subprocess calls, or privilege escalation attempts (such as sudo or chmod) in the skill instructions or reference data.\n- [DYNAMIC_CONTEXT_INJECTION]: No usage of dynamic shell execution placeholders or argument injection patterns was found in the SKILL.md file.\n- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to respond to user queries about screenwriting, it lacks any automated tools (network, file-write, or code execution) that could be exploited through malicious user input, rendering the attack surface non-existent.

sw-series-structure