prd-creation
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The workflow instructions in
references/current-guidance.mdrequire the agent to interpolate untrusted user input into a command-line template for execution. - Ingestion points: The
<USER_REQUEST>placeholder in theMORU_LAUNCHERcommand template inreferences/current-guidance.mdcaptures arbitrary user input. - Boundary markers: The input is encapsulated in double quotes within the shell command, but there are no explicit instructions to the agent to sanitize or ignore instructions embedded within that user text.
- Capability inventory: The skill utilizes command execution via the
<MORU_LAUNCHER>utility and possesses file system read/write capabilities for document generation. - Sanitization: There is no evidence of validation or sanitization logic applied to the user request before it is used as a command argument.
- [COMMAND_EXECUTION]: The instructions explicitly direct the agent to execute a CLI tool (
<MORU_LAUNCHER>) and several local scripts (workflow.py,agent-preflight.py,agent-finish-check.py) to manage the workflow process.
Audit Metadata