ai-search

SUSPICIOUS. The stated purpose matches the behavior: this is a web fact-checking wrapper around official AI CLIs, not a disguised credential stealer. Main risk comes from executing arbitrary AI_SEARCH_CMD through sh -c and forwarding user questions to third-party backends, which is proportionate to the purpose but still materially risky if prompts or environment variables are not controlled.

The code implements an AI-powered web-search workflow and contains no clear malware or sabotage indicators in the supplied fragment. The main security risk is intentional arbitrary shell execution through the environment-controlled AI_SEARCH_CMD, followed by possible disclosure of question and backend data to external services or logs. Filename handling is improved with sanitization and mktemp but should use a strictly safe display encoding if terminal escape injection is a concern.