The Agent Skills Directory

Persistence Mechanisms (HIGH): The setup instructions for scripts/tapi-auth.js explicitly state that the API key is 'saved to your shell profile'. This implies the script modifies sensitive configuration files such as ~/.bashrc, ~/.zshrc, or ~/.profile to inject environment variables. Modifying shell startup scripts is a high-risk persistence technique used to maintain access or alter the execution environment.
Indirect Prompt Injection (LOW): The skill is designed to ingest untrusted data from an external source (YouTube transcripts via a third-party API).
Ingestion points: Data enters through the https://transcriptapi.com/api/v2/youtube/transcript endpoint.
Boundary markers: There are no delimiters or instructions provided to the agent to treat the fetched transcript as untrusted data.
Capability inventory: The agent has the ability to execute node scripts (during setup) and perform network operations via curl.
Sanitization: No evidence of sanitization or filtering of the transcript content exists, allowing potentially malicious instructions embedded in subtitles to influence agent behavior.
Command Execution (MEDIUM): The skill relies on executing local Node.js scripts (scripts/tapi-auth.js) with user-provided inputs (USER_EMAIL, TOKEN_FROM_STEP_1, CODE). While intended for authentication, these scripts execute outside the standard agent context and modify system configuration.

subtitles