Skills
Navigate
skills/whytryharder/superpowers/ai-rag-pipeline/Gen Agent Trust Hub

ai-rag-pipeline

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external web searches and URL extraction and passes it to LLMs.
  • Ingestion points: Variables such as $SEARCH, $CONTENT, $EVIDENCE, $OVERVIEW, and $STATISTICS are populated with untrusted data from web search tools (Tavily, Exa).
  • Boundary markers: Prompt templates use simple labels like 'Search Results:' or 'Research Data:' but lack robust delimiters or instructions to ignore embedded commands.
  • Capability inventory: The skill uses the belt CLI (allowed via frontmatter) to invoke multiple LLMs and tools.
  • Sanitization: No sanitization or validation is applied to external content before prompt interpolation.
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct users to download and install external software.
  • Evidence: Mentions npx skills add belt-sh/cli and an installation script at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md.
  • [COMMAND_EXECUTION]: The skill body contains several shell command examples using the belt tool to manage data flow.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 05:50 AM