steel-manning
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were detected in the skill instructions or metadata.
- [COMMAND_EXECUTION]: The skill utilizes specific MCP tools (
mcp__wiki-vault__vault_search,mcp__wiki-vault__vault_query_graph,mcp__wiki-vault__vault_add_edge) to manage decision context within a wiki environment. These are legitimate tool uses for the stated purpose of tracking and challenging decision-making processes. - [PROMPT_INJECTION]: While the skill uses terminology like 'adversarial verification', 'attack vectors', and 'Red Teaming', these refer to established logical debate and stress-testing methodologies rather than technical attempts to bypass AI safety guardrails.
- [DATA_EXFILTRATION]: No unauthorized network calls or patterns indicating exfiltration of sensitive information were identified. Data interactions are limited to the provided vault tools.
Audit Metadata