The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external files (e.g., store.md, products.md, audience.md) and interpolate this content into prompts for generating marketing copy and visual assets. The instructions lack explicit boundary markers or delimiters to isolate this potentially untrusted external data from the skill's core system instructions.
Ingestion points: The skill reads brand-specific data such as store URLs, target audience profiles, and product details from files within the 'Brand Brain' directory.
Boundary markers: The workflow does not define any specific delimiters (e.g., XML tags or unique markers) to prevent the agent from misinterpreting instructions that might be embedded within the brand data files.
Capability inventory: The skill possesses significant capabilities, including generating complex prompts for AI image generators, producing ad copy for social platforms, and appending new data to the brand memory file (learnings.jsonl).
Sanitization: There is no evidence of sanitization, validation, or escaping logic applied to the external content before it is interpolated into the agent's context or used to drive next-step decisions.

afa-creative