14-red-team-ops
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The
scripts/engagement_planner.pyscript accepts user-provided JSON input to generate a formatted red team engagement plan, which creates a potential surface for indirect prompt injection.\n - Ingestion points:
scripts/engagement_planner.pyreads external JSON data via the--scopecommand-line argument.\n - Boundary markers: The script does not utilize explicit delimiters or safety instructions within the generated Markdown output to isolate user-provided data from agent instructions.\n
- Capability inventory: The skill documents commands for network exploitation, credential harvesting, and lateral movement using tools such as Impacket, BloodHound, and Mimikatz.\n
- Sanitization: No input validation or escaping is applied to the JSON content before it is rendered into the Markdown plan.\n- [SAFE]: The skill uses standard dependencies (
pyyaml,requests) and references reputable security resources and tools. It includes a built-in authorization verification process that prompts the user for engagement details, scope, and points of contact before providing operational assistance.
Audit Metadata