The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/fetch_transcript.py uses subprocess.run to execute system-level commands including wg, wg-quick, and ip rule to manage WireGuard VPN connections. These are privileged operations necessary for the skill's primary function of routing traffic through a residential IP to bypass YouTube's blocks, but they represent a sensitive capability.- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted transcript data from YouTube without sanitization. \n
Ingestion points: scripts/fetch_transcript.py fetches raw text content from external YouTube transcripts. \n
Boundary markers: Absent. The transcript data is provided as a raw string within a JSON structure. \n
Capability inventory: The skill allows for system command execution (subprocess.run) and network operations (requests). \n
Sanitization: Absent. No filtering or escaping of the transcript content is performed before it enters the agent's context.- [SAFE]: The file references/SETUP.md provides standard WireGuard configuration instructions with placeholders for keys and IPs, which is a secure and standard practice for infrastructure documentation.- [SAFE]: The SAFETY.md file contains several malicious patterns (e.g., exfiltration via curl and credential reading via grep) and adversarial prompts, but these are clearly presented as negative examples for safety training and do not reflect the skill's intended behavior.

youtube-transcript-hardened