The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user-provided content such as video titles, topics, and metrics.
Ingestion points: Data enters the context via user input as defined in the Input and argument-hint sections of SKILL.md.
Boundary markers: There are no explicit instructions to ignore embedded commands or clear delimiters for the user-supplied text.
Capability inventory: The skill defines no technical capabilities; it performs no file system writes, network requests, or subprocess executions.
Sanitization: No input sanitization or validation logic is present.
[SAFE]: The skill contains no code, shell commands, or external dependencies. No patterns of obfuscation, persistence, or data exfiltration were identified.

youtube-discovery