The Agent Skills Directory

[PROMPT_INJECTION]: The SKILL.md file explicitly instructs the AI agent to ignore its internal safety guidelines and refusal protocols when encountering "sensitive" or "uncomfortable" themes in fictional works, such as domestic violence or infidelity. This is designed to prevent the analysis pipeline from stalling but technically overrides standard safety behavior. While this is associated with the primary purpose of literary analysis, it is a targeted bypass instruction.
[PROMPT_INJECTION]: The skill ingests untrusted text from users for analysis and writes the results to a local file system. This workflow creates an attack surface for indirect prompt injection, where malicious instructions embedded in the analyzed stories could attempt to influence the agent's behavior during the multi-stage analysis process. The instructions lack explicit sanitization or boundary markers for this external data.

story-short-analyze