scaffolding-openai-agents

Warn

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The calculate function tool in SKILL.md uses the eval() function to process mathematical expressions. Because the input to this tool is generated by the AI agent based on potentially untrusted user prompts, it allows for arbitrary Python code execution. Although a code comment suggests using safe_eval in production, providing the raw eval() implementation remains a significant security risk for developers using the scaffold.
  • [INDIRECT_PROMPT_INJECTION]: The skill architecture is susceptible to indirect prompt injection as it processes untrusted user data that can influence agent tool-calling and responses.
  • Ingestion points: User queries are ingested through the Runner.run method in the main.py example and the FastAPI /ask endpoint in SKILL.md.
  • Boundary markers: The provided agent instructions lack explicit delimiters or specific guard instructions to isolate user data from agent logic.
  • Capability inventory: Tools defined in SKILL.md include powerful capabilities such as code execution via eval() and network operations via httpx.
  • Sanitization: The implemented guardrails in SKILL.md rely on simple keyword matching (e.g., checking for 'python', 'solution') and do not provide robust sanitization against sophisticated prompt injection attempts.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 22, 2026, 07:13 PM