The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data to populate video content, creating a potential surface for indirect prompt injection.
Ingestion points: The AppFactoryDemoProps interface in SKILL.md takes fields such as title and highlights from external sources.
Boundary markers: There are no instructions or delimiters provided to ensure the agent treats these inputs as untrusted content.
Capability inventory: The skill is authorized to use the Bash tool for rendering tasks.
Sanitization: No guidelines are provided for escaping or validating these inputs before they are interpolated into command-line arguments.
[COMMAND_EXECUTION]: The skill workflow involves executing local scripts and established framework binaries via the command line.
Evidence: The manual rendering instructions in SKILL.md use node scripts/render-demo-video.mjs and npx remotion, which are executed in the local environment.

remotion