The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of a local authentication script (scripts/tapi-auth.js) which modifies several user configuration files to store the TRANSCRIPT_API_KEY environment variable.
Persistence Mechanisms: The script modifies shell startup files including ~/.bashrc, ~/.zshenv, ~/.profile, and ~/.zprofile to export the API key for future sessions.
System Configuration: It also modifies PowerShell profiles on Windows, Fish shell configurations, and agent-specific files like ~/.openclaw/openclaw.json.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from YouTube.
Ingestion points: The skill fetches video titles, descriptions, and transcripts from https://transcriptapi.com (which acts as a proxy for YouTube content) into the agent's context.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the fetched YouTube content as untrusted data or to ignore embedded instructions.
Capability inventory: The skill possesses capabilities for network requests and local command execution (via the setup script).
Sanitization: There is no evidence of sanitization or filtering applied to the search results or transcripts before they are processed by the agent.

youtube-search