The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external brand profiles and user-provided angles, which are then interpolated into prompts for image and video generation tools.
Ingestion points: External data enters the agent's context through tools like advibly_get_brand (identity, tone, research brief) and advibly_get_brand_dossier (voice-of-customer lines), as well as direct user input for the ad's "angle" or "idea" in Phase 1.
Boundary markers: The skill implements a mandatory "Phase 2" approval gate where the agent must present a structured concept and beat list (in JSON format) for user review and confirmation before any generation occurs. It also utilizes a "STYLE LOCK" and "NEGATIVE" prompt structure to constrain model outputs.
Capability inventory: The skill has access to a wide range of creative and distribution tools, including advibly_generate_image, advibly_generate_video, advibly_generate_voiceover, advibly_generate_music, advibly_render_composition, advibly_add_subtitles, and advibly_social_create_post.
Sanitization: The instructions do not specify technical sanitization or escaping of the ingested content before it is inserted into the generation prompts, relying instead on the manual approval process to detect malicious content.

stickman-animation