langgraph-workflows
Warn
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The code snippet for a 'Dynamic Agent' defines a tool that uses the
eval()function to process input. - Evidence:
const calculate = tool(async ({ expression }) => eval(expression).toString(), ...inSKILL.md. - Risk: The
eval()function executes any string passed to it as code. In an agentic context, where the inputexpressionis generated by an LLM, an attacker could use prompt injection to trick the agent into executing malicious JavaScript commands instead of mathematical expressions.
Audit Metadata