material-3

Pass

Audited by Gen Agent Trust Hub on May 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The audit feature in SKILL.md creates a surface for indirect prompt injection where malicious instructions embedded in a target file or URL could subvert the agent's behavior.
  • Ingestion points: The audit command (defined in SKILL.md) accepts arbitrary URLs and file paths provided by the user.
  • Boundary markers: The instructions lack explicit delimiters or warnings to the agent to disregard instructions found within the audited content.
  • Capability inventory: The skill makes use of filesystem reading, network access via browser tools, and shell command execution (grep) to perform compliance checks.
  • Sanitization: There are no instructions for sanitizing or escaping the data read from external targets before processing.
  • [COMMAND_EXECUTION]: The skill (in SKILL.md) instructs the agent to execute specific grep patterns on the local filesystem to audit source code for hardcoded colors and Material Design 2 patterns. This is a legitimate utility for the skill's purpose but involves direct interaction with the local environment using user-supplied paths.
Audit Metadata
Risk Level
SAFE
Analyzed
May 23, 2026, 03:53 AM
Security Audit — agent-trust-hub — material-3