ai-rag-pipeline
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
infshCLI via Bash to orchestrate RAG pipelines, including logging in and running various applications. This is the primary intended functionality of the skill. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it fetches content from external web sources and interpolates this untrusted data into LLM prompts.
- Ingestion points: Untrusted data enters the agent context via search and extraction tools such as
tavily/search-assistant,exa/search, andtavily/extract(SKILL.md). - Boundary markers: The skill uses simple text labels like 'Search Results:', 'Source 1 (Tavily):', and '== Overview ==' as delimiters, which may not be robust against sophisticated adversarial instructions hidden in retrieved data.
- Capability inventory: The skill possesses the capability to execute shell commands (
Bash(infsh *)) and interact with external search and LLM APIs. - Sanitization: The provided examples do not demonstrate sanitization or escaping of the retrieved content before it is processed by the LLM.
Audit Metadata