carbon-footprint-tracking
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements standard mathematical models for Scope 1, 2, and 3 emissions calculation. The code is transparent, logically sound, and restricted to data processing and reporting tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for ingesting external operational and supplier data. 1. Ingestion points: Data is accepted through input dictionaries in functions like calculate_scope1_facilities, calculate_scope3_transportation, and add_supplier in SKILL.md. 2. Boundary markers: No specific boundary markers are used in the provided Python code snippets to separate data from instructions. 3. Capability inventory: Data manipulation and visualization capabilities; documentation mentions network-capable libraries (requests, beautifulsoup4) for data gathering. 4. Sanitization: Input data is used directly in mathematical calculations without explicit filtering or escaping.
- [SAFE]: No obfuscation, prompt injection, persistence mechanisms, or hardcoded credentials were found in the skill metadata or instructions.
Audit Metadata