spoon-agent-development

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download and execution of external tools using the Model Context Protocol (MCP). It references using npx and uvx to fetch packages at runtime, such as the tavily-mcp tool for web searches.\n- [COMMAND_EXECUTION]: The MCPTool class is designed to execute system commands to run MCP servers. This includes spawning subprocesses for Node.js, Python, and UVX environments as described in the configuration documentation.\n- [PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection because agents are designed to ingest and process external data within a ReAct loop.\n
  • Ingestion points: Data enters the agent via agent.run() and tool outputs from tavily-search in scripts/mcp_agent.py.\n
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands were found in the system prompt templates in references/prompts.md.\n
  • Capability inventory: The skill can execute arbitrary commands via MCPTool and perform network operations via SSE/WebSocket transports.\n
  • Sanitization: No evidence of automated sanitization or filtering of external tool outputs was detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 07:33 PM