decision-debate
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured, multi-persona debate workflow for decision support, requiring the AI to research from independent lenses before synthesizing a recommendation for the user.
- [SAFE]: The skill contains strong negative constraints, explicitly forbidding the agent from modifying files, editing the wiki, or committing changes based on the debate results, ensuring that the final decision remains with the human user.
- [DATA_EXPOSURE]: The skill accesses local resources such as 'agent_registry.md' and the 'MEL Wiki' for research purposes. These operations are within the scope of its defined functionality and do not involve access to sensitive credentials or private system files.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data via WebSearch. While this represents a potential surface for indirect prompt injection, the risk is minimized by the skill's design, which limits its capabilities to synthesis and reporting rather than code execution or persistent file modifications.
Audit Metadata