experiment-guardrail-linter
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, which creates a potential surface for indirect prompt injection attacks.\n
- Ingestion points: The
run.pyscript reads JSON or CSV data from a file path provided through the--inputcommand-line argument.\n - Boundary markers: Absent; the instructions in
SKILL.mddo not specify delimiters or warnings to prevent the agent from following instructions potentially embedded within the input data.\n - Capability inventory: The
run.pyscript has the capability to write files to the local file system using the--outputargument.\n - Sanitization: Absent; the Python implementation does not include validation or sanitization routines for the data before processing and writing the output.
Audit Metadata