The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (video titles, descriptions, and transcripts) from YouTube, creating a potential surface for indirect prompt injection attacks.
Ingestion points: Data returned from all https://transcriptapi.com/api/v2/youtube/channel/ endpoints.
Boundary markers: Not present in the provided instructions.
Capability inventory: Shell command execution via curl and file system operations (read/write/delete) as defined in references/auth-setup.md.
Sanitization: No sanitization or validation of the external content is described.
[PERSISTENCE]: The references/auth-setup.md documentation guides the agent to store the TRANSCRIPT_API_KEY in shell profile files or environment configuration files to ensure the key persists across future sessions. This is used for standard configuration management of the required service credentials.
[COMMAND_EXECUTION]: The skill relies on curl commands to make HTTP requests to the TranscriptAPI service. Ready-to-run examples are provided in SKILL.md and references/curl-examples.md.
[DATA_EXFILTRATION]: The skill performs network requests to transcriptapi.com to fetch video and channel data. While necessary for the skill's primary functionality, the domain is not on the default whitelist for network operations.

youtube-channels