ninja-patterns
Fail
Audited by Gen Agent Trust Hub on Feb 13, 2026
Risk Level: HIGHDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION] (HIGH): Exposure of sensitive local file system paths. The file SKILL.md contains hardcoded absolute paths:
/Users/lgbarn/Personal/Indicators/Ninjatrader/PEMA.csand/Users/lgbarn/Personal/Indicators/Ninjatrader/. This reveals the local username 'lgbarn' and internal directory structure, providing reconnaissance data that could be exploited if the agent has file-system access. - [PROMPT_INJECTION] (LOW): Vulnerability surface for Indirect Prompt Injection (Category 8). 1. Ingestion points: External data retrieved via 'doc-researcher agent' or 'Ref MCP tools'. 2. Boundary markers: Absent in the provided scaffold. 3. Capability inventory: Generation of NinjaScript/C# indicator code and API verification. 4. Sanitization: Absent. There is a risk that malicious external documentation could influence the logic or security of generated trading indicators.
Recommendations
- AI detected serious security threats
Audit Metadata