exposure-risk-quantification
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external 'recon findings' and 'scan.db' data, which are untrusted sources. This provides a surface for attackers to embed malicious instructions within the findings to manipulate the agent's risk quantification or reporting behavior.
- Ingestion points: The skill ingests data from a 'completed set of recon findings' and 'findings already collected' (SKILL.md §0, §1).
- Boundary markers: While the skill provides logic for 'honesty guardrails' and 'demotion caps' to ensure statistical accuracy (SKILL.md §7.5, §12), it does not specify explicit boundary markers or instructions to ignore potential natural language commands embedded within the finding descriptions.
- Capability inventory: The skill instructions refer to rendering reports via 'headless Chromium via Playwright' (SKILL.md §10.5), which represents a high-capability tool environment if exposed to untrusted input.
- Sanitization: There is no evidence of sanitization, filtering, or escaping of the textual content within findings before they are processed by the agent.
Audit Metadata