ai-rag-pipeline
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to ingest untrusted data from external web searches and interpolate it directly into LLM prompts. This creates a surface for indirect prompt injection where malicious content on a searched website could attempt to influence the agent's behavior.
- Ingestion points: Web search results and page content extractions entering the context via Tavily and Exa tools in SKILL.md.
- Boundary markers: None present; data is interpolated using direct variable substitution in prompt strings.
- Capability inventory: The skill uses the Bash tool to execute infsh commands, which can trigger further network operations and tool executions.
- Sanitization: No evidence of sanitization, filtering, or validation of the retrieved web content before it is added to the prompt context.
- [COMMAND_EXECUTION]: The skill provides numerous examples of shell command execution using the infsh utility to run AI applications, manage logins, and process data streams.
- [EXTERNAL_DOWNLOADS]: The documentation references an external installation script for the CLI tool hosted on the vendor's GitHub repository.
Audit Metadata