seeds-regenerative-inquiry-cycle
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security threats identified. The skill is entirely text-based and instructional.
- [PROMPT_INJECTION]: The skill uses strong directives to ensure the AI follows the specific SEEDS framework steps. This is standard behavioral conditioning for an agent and does not attempt to bypass core safety filters or extract system prompts.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data for 'learning_group' and 'place_context' (Ingestion points: SKILL.md input_schema). There are no specific boundary markers present to isolate user input. The skill possesses no dangerous capabilities such as subprocess calls, network operations, or file-writing (Capability inventory: None). No automated sanitization is present. Given the lack of executable capabilities, the risk of indirect injection is negligible.
- [DATA_EXPOSURE]: No hardcoded credentials, API keys, or sensitive file paths were found. The skill does not perform network exfiltration.
Audit Metadata