video-publisher
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its workflow involves processing data from external sources that could contain malicious instructions designed to influence the agent's behavior during metadata generation or the upload process.\n
- Ingestion points: The skill relies on output from
youtube-research-video-topicandyoutube-video-analystto inform its metadata generation, as specified in the 'Content Planning' and 'Analytics' sections.\n - Boundary markers: There are no explicit delimiters or instructions provided in the SKILL.md to treat input from research skills as untrusted or to ignore embedded commands.\n
- Capability inventory: The orchestrator delegates to the
youtube-uploaderskill, which possesses the capability to perform network-based write operations (uploading content and metadata to platform APIs).\n - Sanitization: The skill lacks mechanisms for validating or escaping external content before it is used to generate titles, descriptions, and tags.\n- [NO_CODE]: The skill consists entirely of Markdown-based instructions, workflow diagrams, and configuration tables. It does not include any executable scripts (.py, .js, .sh), binaries, or automated command execution logic.
Audit Metadata