linkedin-content

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill utilizes the tavily/search-assistant tool to research viral LinkedIn patterns from the web. This process involves ingesting untrusted third-party content into the agent's context.
  • Ingestion points: The SKILL.md file defines a research workflow using belt app run tavily/search-assistant to query external web data.
  • Boundary markers: No explicit delimiters or system instructions are provided to the agent to treat the search results as untrusted or to ignore embedded instructions within them.
  • Capability inventory: The skill is granted Bash(belt *) access, allowing it to execute CLI tools, manage logins, and trigger other applications like x/post-create and infsh/html-to-image.
  • Sanitization: There is no evidence of content sanitization or filtering applied to the search results before they are processed by the agent for pattern analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 02:19 PM
Security Audit — agent-trust-hub — linkedin-content