iot-witness-verify

Warn

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx to fetch the package @claude-flow/plugin-iot-cognitum@1.0.0 from the npm registry at runtime.
  • [REMOTE_CODE_EXECUTION]: The package downloaded via npx is executed within the shell environment. This package originates from a third-party namespace that is not categorized as a trusted organization or well-known service.
  • [COMMAND_EXECUTION]: The skill invokes the Bash tool to execute npx and the cognitum-iot CLI tool.
  • [INDIRECT_PROMPT_INJECTION]: The skill accepts a DEVICE_ID argument which is interpolated directly into a shell command and uses tool outputs for decision-making and storage, creating a potential surface for injection if inputs are malicious.
  • Ingestion points: User-provided DEVICE_ID argument in SKILL.md.
  • Boundary markers: None present to delimit user input from the command string.
  • Capability inventory: Shell command execution via the Bash tool and persistent storage via memory_store.
  • Sanitization: No explicit sanitization or validation of the DEVICE_ID is performed before interpolation into the npx command string.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 29, 2026, 09:18 PM