The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-supplied character identity details, style preferences, and reference images, creating a surface where malicious instructions could potentially be embedded in external data sources to influence generation outputs.
Ingestion points: Character identity anchor fields (age, face shape, marks) and external references (source images, style boards) are collected from users as described in SKILL.md.
Boundary markers: The skill uses structured text blocks labeled CHARACTER ANCHOR and SHOT VARIABLE in SKILL.md and references/anchor-system.md to delimit character data, though it does not explicitly instruct the agent to ignore instructions embedded within these fields.
Capability inventory: The skill utilizes the vg generate tool to interact with various image and video generation endpoints from providers including OpenAI, Fal.ai, Bytedance, xAI, and Veed.
Sanitization: No explicit sanitization or validation logic for user-supplied strings or reference content is specified within the skill files.

character-design