The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes methods for reading content from local files (such as brand.md or prompt files) and interpolating that content directly into generative prompts, creating a surface for indirect instructions to influence agent output.
Ingestion points: The skill utilizes data from files via shell commands like cat brand.md in multiple automation examples.
Boundary markers: The examples do not include delimiters or instructions for the AI to ignore embedded commands within the ingested files.
Capability inventory: The skill has the capability to perform network requests and file operations through the Picsart gen-ai CLI.
Sanitization: No sanitization or verification of the external file content is described before inclusion in the prompt string.
[COMMAND_EXECUTION]: The skill documentation includes extensive shell command usage for automation, including piping CLI output into utilities like jq and curl. This is consistent with the skill's purpose as a developer tool for automation.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the @picsart/gen-ai package via npm. As this is a scoped package belonging to the identified author/vendor (picsart), it is considered a legitimate resource.

gen-ai-batch