The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted image data and character references to generate revised prompts, creating a surface for indirect prompt injection.
Ingestion points: Processes user-provided images (as described in SKILL.md Step 1 and 2) and character anchor data from story-character-sheet.
Boundary markers: No explicit delimiters or boundary markers are used to isolate untrusted visual content or character metadata.
Capability inventory: The skill is limited to generating text prompts for future image creation; it does not possess system command execution or network capabilities.
Sanitization: No sanitization or validation mechanisms are described for the visual data or character sheet inputs.
[DATA_EXFILTRATION]: The skill references a local configuration directory to check tool availability.
Evidence: references/rubric-bridge.md references the file path .moai/config/sections/design.yaml to verify the execution environment for the design-iteration-loop tool.

story-webtoon-qc