Skills
skills/qu-skills/skills/python-sdk/Gen Agent Trust Hub

python-sdk

Pass

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The SDK provides a framework for building agents that ingest untrusted data from users or external sources.\n
  • Ingestion points: Untrusted data enters the agent context via agent.send_message and app input parameters in SKILL.md.\n
  • Boundary markers: The documentation encourages the use of system prompts and structured tool schemas to define operational boundaries for agents.\n
  • Capability inventory: The SDK allows for powerful capabilities such as code execution (code_execution(True)) and webhooks (webhook_tool) as shown in references/tool-builder.md.\n
  • Sanitization: The documentation emphasizes the use of require_approval() to implement human review checkpoints for sensitive operations.\n- [COMMAND_EXECUTION]: A tutorial example in references/tool-builder.md demonstrates implementing a calculator tool using the Python eval() function. This is a common instructional pattern but represents a potential risk if implemented exactly as shown with untrusted input without proper validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 16, 2026, 10:49 PM
Security Audit — agent-trust-hub — python-sdk