calorie-counter

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is purely instructional and relies on the AI agent's native capabilities to parse user input and manage a local JSON file.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: Data storage is restricted to a local path (~/Documents/xiaodazi/calories.json). No network requests or data transmission logic is present.
  • [REMOTE_CODE_EXECUTION]: There are no scripts (.sh, .py, .js) or commands that download or execute external code.
  • [INDIRECT_PROMPT_INJECTION]: While the skill processes untrusted user input (food descriptions), it has no high-risk capabilities (like shell access or network requests) to exploit.
  • Ingestion points: User food descriptions provided via chat.
  • Boundary markers: Not explicitly defined.
  • Capability inventory: Local file write to a specific JSON path.
  • Sanitization: None, but the risk is minimal given the lack of dangerous tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 04:19 PM