record-episode
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script
harness/scripts/mem-rank.pythrough the system shell. This is an intended function but creates a potential execution vector for malicious commands if the inputs are not strictly controlled. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating potentially untrusted user data directly into shell commands.
- Ingestion points: The skill takes input for
did(session summary),outcome(result), andsession(identifier) inSKILL.md. - Boundary markers: No explicit delimiters or boundary markers are defined to prevent the agent from misinterpreting instructions embedded within these data fields.
- Capability inventory: The skill uses shell execution to run the Python interpreter on
harness/scripts/mem-rank.py(documented inSKILL.md). - Sanitization: There are no instructions or mechanisms provided to sanitize, validate, or escape the user-supplied strings before they are interpolated into the shell command string.
Audit Metadata