managing-detox

Warn

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple system and project-level CLI tools to discover configurations and available devices. Evidence: Use of npx detox, xcrun simctl, emulator, find, and tail in bash scripts.
  • [REMOTE_CODE_EXECUTION]: The skill uses dynamic execution to load and evaluate local JavaScript configuration files. Evidence: The detox_config helper function executes node -e "require('./.detoxrc.js')" and node -e "require('./detox.config.js')", which runs the code contained within those files at runtime.
  • [PROMPT_INJECTION]: The skill processes untrusted test artifacts, creating a surface for indirect prompt injection. Ingestion points: Reads contents of *.log files from the artifacts/ directory in Phase 2. Boundary markers: Absent; logs are directly tailed into the output without delimiters or instructions to ignore embedded commands. Capability inventory: The skill has the ability to execute shell commands (npx, node, find, etc.). Sanitization: Absent; no filtering or validation is performed on log content before presentation to the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 23, 2026, 11:48 AM
Security Audit — agent-trust-hub — managing-detox