run-aql
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documents shell command templates for the
anfraCLI utility to execute data queries. It utilizes placeholders for dataset names and query strings. - [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted query data through a command-line interface.
- Ingestion points: Untrusted data enters the execution context through the
<query>and<uname>placeholders inSKILL.md. - Boundary markers: The documentation suggests wrapping queries in single quotes within shell commands to mitigate basic shell injection.
- Capability inventory: The skill facilitates shell command execution and database interaction via the
anfratool. - Sanitization: The skill instructions explicitly direct the agent to use a separate
validate-aqlprocess to check queries before execution.
Audit Metadata