The Agent Skills Directory

[SAFE]: The skill serves as an instructional framework for narrative theory distillation and does not include any executable scripts, network operations, or sensitive file system access.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external narrative sources (texts, interviews, production documents) as part of its core function. While this is an ingestion surface for untrusted data, the skill lacks access to sensitive tools (network, file-write, or shell access) that could be exploited via an injection. The findings are as follows:
Ingestion points: External storytelling sources identified in SKILL.md (Workflow Step 1).
Boundary markers: The EXTRACT protocol and AXIOM_CRITERIA provide structural delimiters for the analysis output, though explicit delimiters for the input data itself are not defined.
Capability inventory: Limited to text generation, formalization (pseudocode/tables), and diagnostic question generation. No system-level tools are invoked.
Sanitization: None present, which is typical for text-to-text transformation skills.
[SAFE]: The inclusion of Sanskrit, Japanese, and Chinese terminology in references/theoretical-correspondences.md consists of legitimate narratological terms (e.g., Rasa, Yũgen, Wen). These are not homoglyph attacks or obfuscation attempts.
[SAFE]: Relative file paths and external documentation links (e.g., to creative-coding-skills-ecosystem.md) are standard navigational references within a repository and do not target sensitive system locations.

narratological-algorithms