The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a custom CLI, genmedia, to manage the asset production lifecycle. Commands like genmedia models, genmedia upload, and genmedia run are employed to interact with remote inference endpoints. This is consistent with the skill's objective of generating marketing media.
[EXTERNAL_DOWNLOADS]: The skill performs automated downloads of generated content from external services to the local ./outputs/marketing/ directory. These downloads are initiated by the genmedia run and genmedia status tools using a user-specified path template.
[INDIRECT_PROMPT_INJECTION]: The skill ingests various external inputs, such as user objectives and product claims, which are then used to build prompts for generative models.
Ingestion points: User-provided marketing details collected in SKILL.md.
Boundary markers: None explicitly defined for input interpolation.
Capability inventory: Local file reading (genmedia upload), file writing (genmedia run --download), and network requests (CLI model calls) as documented in SKILL.md.
Sanitization: The skill includes 'Claim discipline' and 'Guardrails' sections in SKILL.md and references/campaign-patterns.md to restrict the model from generating fake or harmful content.

marketing