The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill takes multiple user inputs and interpolates them directly into prompts for image generation models, creating a vulnerability surface where malicious input could influence the generated content.
Ingestion points: User-provided inputs defined in SKILL.md, specifically product_name, product_category, key_features, and target_buyer.
Boundary markers: Absent. The skill places user-supplied strings directly into natural language prompts without delimiters (like triple quotes or XML tags) or instructions to ignore embedded commands.
Capability inventory: The skill uses the muapi CLI and curl to make network requests to an external API for image generation.
Sanitization: Absent. There is no evidence of input validation, filtering, or escaping of the user-provided strings before they are sent to the AI models.
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands to interact with the MuAPI service.
Evidence: The instructions in SKILL.md prompt the agent to call muapi image generate, muapi auth configure, and curl -X POST https://api.muapi.ai/... using the user's API key.
[EXTERNAL_DOWNLOADS]: The skill interacts with an external API service to perform its primary image generation tasks.
Evidence: The skill fetches generated content from https://api.muapi.ai/api/v1/ based on user inputs. This is the intended primary purpose of the skill and uses vendor-owned infrastructure.

muapi-amazon-product-listing