skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/data-evolution-analysis/Gen Agent Trust Hub
data-evolution-analysis
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The instructions do not contain any patterns attempting to override agent behavior, bypass safety guidelines, or extract system prompts.
- [DATA_EXFILTRATION]: No network operations (curl, wget, or HTTP requests) are present. There is no evidence of hardcoded credentials or access to sensitive system file paths.
- [REMOTE_CODE_EXECUTION]: The skill uses standard Python libraries (json, dataclasses, enum) and does not download or execute remote scripts or packages.
- [COMMAND_EXECUTION]: No dangerous system commands, subprocess spawning, or privilege escalation attempts (like sudo or chmod) were detected.
- [DATA_EXPOSURE]: While the skill requests filesystem permissions via its configuration, this is aligned with its stated purpose of reading project data and writing maturity reports. It does not target sensitive directories.
- [PROMPT_INJECTION]: Indirect injection risk is low as the skill processes structured data (CSV, JSON, Excel) and includes instructions for the agent to validate all inputs before processing.
Audit Metadata