gtm-account-scoring

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates entirely on locally provided facts and context files located in a specific local directory structure (~/.gtm/). It explicitly prohibits external system changes, web access, or link opening to prevent data leakage.
  • [DATA_EXFILTRATION]: The skill includes explicit security instructions in references/context.md to never echo or retain credentials, tokens, or keys that might be embedded in the input data, mitigating potential credential exposure during the scoring process.
  • [COMMAND_EXECUTION]: While the skill interacts with the filesystem to resolve GTM contexts, it does so using predefined relative paths and does not utilize arbitrary shell commands or user-injected arguments for execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted account data and ICP prose. It mitigates injection risks by defining strict output schemas (literal item fields) and enforcing a 'supplied-facts-only' boundary that prevents the agent from following instructions contained within the analyzed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 03:45 PM
Security Audit — agent-trust-hub — gtm-account-scoring