domain-list
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
Bashtool and incorporates user-provided$ARGUMENTSfor filtering operations. This creates a surface for argument injection if the input is not correctly validated before being passed to the shell. - [PROMPT_INJECTION]: The skill processes data from
.workflow/domain/glossary.yaml, which represents an indirect prompt injection surface. - Ingestion points: Data enters the agent context from the file
.workflow/domain/glossary.yamlduring Step 1 of the execution. - Boundary markers: The skill includes an 'Output boundary' context section instructing the agent that the command is read-only and must only display output to the console.
- Capability inventory: The skill is permitted to use
Bash,Read,Glob, andGreptools. - Sanitization: No explicit sanitization or validation of the glossary content or the
$ARGUMENTSinput is specified in the skill instructions.
Audit Metadata