The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from users, which creates a standard surface for indirect prompt injection where instructions embedded in source assets might influence downstream AI model behavior.
Ingestion points: Defined in SKILL.md (source assets like transcripts and brand guides) and references/node-rules.md (the 'extractor' node which converts assets into prompts).
Boundary markers: The skill does not explicitly instruct the agent to use delimiters or specific safety warnings to prevent the agent from obeying instructions found within user-provided assets.
Capability inventory: The skill utilizes the genmedia CLI to perform network-based API calls (genmedia run) and write output files to the local disk (genmedia status --download).
Sanitization: No explicit instructions are provided for sanitizing or filtering user-provided text before it is interpolated into prompts for generation nodes.
[COMMAND_EXECUTION]: The skill's primary functionality is built around the execution of the genmedia CLI. These commands are used for model discovery, schema inspection, asset uploading, and media generation.
The genmedia CLI is a tool associated with the skill's authoring community, and its use is the primary stated purpose of the skill.

genmedia-workflow