x-posting
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from blog posts and podcasts to generate social media posts. While it includes a human approval step, the instructions lack explicit boundary markers or content sanitization, creating a surface for indirect prompt injection from the source material.
- Ingestion points: The skill reads file paths and content from
Content/Master Content Database/Podcasts/andContent/Master Content Database/Blog Posts/to extract insights and quotes. - Boundary markers: The prompt template used to spawn sub-agents does not utilize delimiters or specific instructions to ignore embedded commands within the ingested source text.
- Capability inventory: The skill is capable of executing shell commands via subprocess calls and performing network operations to external APIs.
- Sanitization: No explicit sanitization or filtering of the source content is documented before interpolation into agent prompts.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command using string interpolation for user-provided content:
python3 agents/post_tweet.py '[tweet content]'. This pattern is vulnerable to command injection if the content contains shell metacharacters or unescaped quotes. - [DATA_EXFILTRATION]: The skill performs network operations to
getlate.devto post and schedule content. While this is the intended purpose and follows best practices by retrieving credentials from an environment file, it involves sending data to a third-party domain not on the standard whitelist.
Audit Metadata