nth-event-tag-builder
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing a sequence of shell commands via the
ae-cliutility to interact with the target platform's metadata and analysis configurations.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input—including project identifiers, event names, and user properties—and interpolates these values into shell commands and SQL query templates.\n - Ingestion points: User responses to mandatory configuration questions defined in the
SKILL.mdworkflow.\n - Boundary markers: There are no specialized delimiters or 'ignore' instructions used to wrap user-provided variables within the generated command strings or SQL blocks.\n
- Capability inventory: The skill uses the
ae-clitool to create persistent entities, virtual properties, and SQL tags on a remote project, which could be exploited if the inputs are manipulated.\n - Sanitization: The skill suggests using explicit type-casting for numeric identifiers in SQL (e.g.,
cast("<userID>" as varchar)) to prevent concatenation errors, which provides minimal type-level protection but does not prevent logic-based injection attacks.
Audit Metadata