The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides templates that incorporate untrusted user data—both text prompts and image URLs—into API requests, which could lead to unintended behavior if the underlying model processes malicious instructions within that data.
Ingestion points: User-provided text content and image URLs in the messages array of the API request examples (SKILL.md).
Boundary markers: The examples do not include explicit delimiters or instructions to ignore potential commands embedded within the user-supplied content.
Capability inventory: The skill is configured to use Bash(curl *) for network-based API interactions.
Sanitization: No sanitization, escaping, or validation logic for the external inputs is described or implemented in the provided examples.
[EXTERNAL_DOWNLOADS]: The documentation includes examples of fetching remote assets via URLs (e.g., https://example.com/channel-banner.jpg) and establishes communication with the external API at https://eachsense-agent.core.eachlabs.run. This is the core functionality for processing and generating visual content.
[COMMAND_EXECUTION]: The skill provides multiple curl command structures intended to be executed in a shell environment to interact with the EachLabs service. These commands include the use of environment variables for API key management and JSON payloads for request configuration.

youtube-content