podcast-script-generator
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes external content from user-specified local files or direct text input without using boundary markers or sanitization techniques.
- Ingestion points: Content is ingested via the Read tool from local markdown files or provided text in Step 1 of the workflow.
- Boundary markers: There are no specific delimiters or instructions for the agent to ignore potentially malicious instructions embedded within the source podcast notes.
- Capability inventory: The skill is primarily focused on text analysis and transformation using a Read tool; it does not invoke high-risk execution tools like shells or interpreters on the ingested data.
- Sanitization: The skill does not implement any filtering or validation of the input content before processing it.
- [DATA_EXFILTRATION]: The skill instructions include a hardcoded local file path (
/Users/ugreen/Documents/obsidian/每日播客/) to facilitate reading notes from an Obsidian vault. While this represents access to personal user documents, it is functionally consistent with the skill's purpose as a podcast script generator. No mechanisms for exfiltrating this data to external servers were found.
Audit Metadata