component-tester

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The script scripts/test-memex-cli.sh references memex-cli, an external Node.js package not listed in the trusted sources. Its suggested installation via npm constitutes an unverifiable dependency.
  • [PROMPT_INJECTION] (MEDIUM): The memex-cli tool serves as an interface for LLM backends, presenting a surface for indirect prompt injection. * Ingestion points: The --prompt argument in scripts/test-memex-cli.sh. * Boundary markers: None implemented in the test script call. * Capability inventory: Network communication with AI providers (Anthropic, OpenAI, Google) and local command execution. * Sanitization: No sanitization or filtering of prompt content is observed.
  • [COMMAND_EXECUTION] (LOW): The test script executes shell commands and the memex-cli binary to verify system status and functionality.
  • [CREDENTIALS_UNSAFE] (LOW): The script checks for the presence of several sensitive API keys (ANTHROPIC_API_KEY, OPENAI_API_KEY, GOOGLE_API_KEY, GEMINI_API_KEY) in the environment. Although the keys are not exfiltrated, identifying their presence is a security-relevant action.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 01:21 PM