The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation includes instructions for managing modular dependencies via the npx skills add command. This allows the agent to install necessary 'sibling skills' from the scenario-labs/skills repository to handle specific parts of the production pipeline, such as audio generation or video assembly. As these resources are hosted in the official repository of the skill's author, this represents legitimate cross-skill coordination and dependency management.
Evidence: references/dependencies.md explicitly lists commands like npx skills add scenario-labs/skills --skill <name> to resolve gaps in the available toolset.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to process external assets such as product photos, logos, and creative briefs provided by the user or external sources.
Ingestion points: references/brief.md defines product shots, character references, and font files as external inputs required to start a campaign.
Capability inventory: The agent uses tools for image-to-video generation, audio synthesis, and final video assembly.
Sanitization: The skill mitigates risks associated with untrusted data by implementing a 'Fidelity Gate' (Step 6), which uses the asset_analyze tool to perform automated visual validation of every generated clip against the original, approved brand reference before it is allowed into the final edit.
[COMMAND_EXECUTION]: The instructions direct the agent to utilize specialized technical tools via search-based discovery for post-production tasks.
Evidence: references/delivery.md and references/storyboard.md mention performing search queries for video upscale and video reverse tools to conform assets to delivery standards.

scenario-video-ads