python-sdk
Pass
Audited by Gen Agent Trust Hub on May 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Documentation examples in 'references/tool-builder.md' and 'references/agent-patterns.md' demonstrate the use of the 'eval()' function to process expressions provided as tool arguments. This is an unsafe coding practice when dealing with data generated by an AI agent or supplied by a user, as it can be exploited for arbitrary code execution if the input contains malicious instructions.
- [EXTERNAL_DOWNLOADS]: The skill instructs the user and agent to install the 'inferencesh' Python package from an external registry. This is the official vendor-provided SDK for the 'inference.sh' platform.
- [PROMPT_INJECTION]: The SDK enables the creation of agents that process untrusted external data (via 'agent.send_message' or tool results) and have access to high-privilege tools such as network requests ('webhook_tool') and code execution. This architecture constitutes a surface for indirect prompt injection, where an attacker-controlled input could attempt to hijack the agent's logic flow.
- [SAFE]: The skill promotes secure authentication practices by recommending the use of environment variables ('INFERENCE_API_KEY') for secret management instead of hardcoding credentials.
Audit Metadata