model-compare

SUSPICIOUS. The main comparison workflow is coherent and mostly uses official provider APIs, but Hybrid Mode expands scope by reading a local .env secret and forwarding it to external CLI tools, with sandbox-disabled execution guidance. That credential-handling and external-CLI orchestration are disproportionate to the otherwise straightforward model-comparison purpose.

No explicit obfuscated malware or hardcoded malicious endpoints are evident in this module. However, the “tools” mode implements a high-impact remote capability: it executes LLM-provided Python via subprocess (run_python) with only a temporary working directory and timeout, and it fetches attacker-controlled URLs via web_extract without allowlisting. Combined with automatic loading of API keys into environment variables, this design can enable secret theft/exfiltration and SSRF/internal probing if an adversary can influence prompts or tool_call outputs. Treat this package as high security risk; isolate strongly and avoid using tool-calling with untrusted inputs.