leadership-sync
Warn
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill is configured to read from sensitive internal files including 'data/cfo/latest_forecast.json' (financial projections), 'data/cfo/assumptions.json' (business model parameters), and 'data/cfo/cap_table.json' (equity structure). This represents a data exposure surface for highly sensitive organizational information.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by consuming data from various external-facing files like 'data/gtm/deal_intel_summary.json' and 'data/product/competitive_analysis.json'. 1. Ingestion points: Data is pulled from multiple JSON files across four departmental layers. 2. Boundary markers: No delimiters are specified to isolate data from agent instructions. 3. Capability inventory: The skill directs the agent to interact with other specialized functional skills (e.g., /cmo, /cto) based on the analyzed data. 4. Sanitization: There is no evidence of data validation or input cleaning before the information is processed by the model.
Audit Metadata