The Agent Skills Directory

[PROMPT_INJECTION]: Analysis of the skill instructions found no attempts to bypass safety filters or override agent behavior. The instructions are focused on technical deployment and guidance for literary analysis and writing workflows.
[DATA_EXFILTRATION]: No hardcoded credentials or data exfiltration patterns were detected. The skill accesses local project files related to writing (outlines, settings, tracking) and includes a version check that calls the vendor's official GitHub repository.
[OBFUSCATION]: No obfuscated code, Base64 commands, or hidden characters were found in the provided files.
[REMOTE_CODE_EXECUTION]: The skill uses internal Python and Node.js scripts for materializing the environment and merging configurations. It does not download or execute untrusted remote scripts.
[PRIVILEGE_ESCALATION]: The skill does not request administrative privileges or use dangerous commands like sudo or chmod 777. It correctly manages executable bits for its own hook scripts.
[PERSISTENCE]: The skill configures a git pre-commit hook to provide advisory warnings during the writing process, which is a documented and legitimate part of the writing toolkit's functionality.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external text for analysis and research purposes. It employs specialized guard scripts (e.g., analysis-input-guard) to enforce writing boundaries and validate input, effectively mitigating risk for this category.
[DYNAMIC_CONTEXT_INJECTION]: No dynamic context injection patterns using the ! syntax were found in the SKILL.md file.

story-setup