youtube-content-gap-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a shell script (
scripts/crawlora.sh) to interface with the Crawlora API. The script implements significant security controls, including strict character validation for API keys, allow-listing of specific API paths via regex, and mandatory URL-encoding of all query parameters. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including YouTube video metadata, search results, and transcript content. This constitutes an attack surface for indirect prompt injection.
- Ingestion points: YouTube metadata and transcripts are ingested via the Crawlora API through
scripts/crawlora.sh. - Boundary markers: There are no explicit instructions for the agent to use delimiters or ignore embedded instructions when analyzing transcript text.
- Capability inventory: The skill allows for network requests via the helper script and file system operations for report generation.
- Sanitization: While the script validates its inputs, the content returned from the external API is processed without specific sanitization filters.
- [SAFE]: API requests are exclusively directed to the vendor's official domain (
api.crawlora.net). - [SAFE]: The implementation follows secure credential handling practices by using temporary
curlconfiguration files to transmit API keys, preventing their exposure in process lists.
Audit Metadata