well-architected-framework-assessment
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows its stated purpose of performing an architecture review. All operations are contained within the user's local environment and Snowflake account.
- [COMMAND_EXECUTION]: The skill uses
snowflake_sql_executeto run diagnostic queries against theSNOWFLAKE.ACCOUNT_USAGEschema. These queries are statically defined in the skill's reference files and are used to calculate the assessment metrics. - [PRIVILEGE_ESCALATION]: The skill explicitly requests use of the
ACCOUNTADMINrole. This high privilege level is a legitimate requirement for accessing the account-wide metadata views necessary for a full WAF assessment. - [DATA_EXFILTRATION]: No network activity to external or non-whitelisted domains was observed. Data gathered during the assessment is used to generate an inline report and an optional local HTML file at a location specified by the user.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests metadata from the Snowflake account, the risk of indirect prompt injection is low as the data processed is primarily numeric or configuration metrics. The skill implements boundary markers via structured JSON output and sanitizes identifiers before use in file operations.
Audit Metadata