youtube-data-api-expert
Warn
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [DYNAMIC_EXECUTION]: The documentation in 'references/auth.md' provides a code snippet that uses 'pickle.load()' to deserialize authentication credentials from a local file named 'token.pickle'. Deserializing untrusted data with 'pickle' is unsafe and can lead to arbitrary code execution if the file is compromised.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on how to retrieve untrusted data from YouTube, such as comments and video metadata, as seen in 'references/comments.md' and 'references/videos.md'. 1. Ingestion points: Data enters the context via 'commentThreads().list()', 'comments().list()', and 'videos().list()'. 2. Boundary markers: The instructions lack delimiters or warnings to ignore embedded instructions in the retrieved text. 3. Capability inventory: The skill enables the agent to perform authenticated actions such as uploading videos and posting comments. 4. Sanitization: There is no evidence of sanitization or validation of the retrieved content.
- [EXTERNAL_DOWNLOADS]: The skill recommends installing official libraries like 'google-api-python-client' and 'google-auth-oauthlib' from trusted sources.
Audit Metadata