problem-analysis

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions and agent roles are consistent with the stated purpose of business problem diagnosis. The multi-agent workflow (tree-builder, external-check, hypothesis, data-mapper, verdict, and critic agents) ensures structured analysis with clear separation of concerns and quality gates.
  • [SAFE]: External tool usage (WebSearch, WebFetch) is appropriately scoped to gathering market benchmarks and competitor information, aligning with the primary strategic purpose of the skill.
  • [SAFE]: The skill processes user-supplied data such as CSVs and reports (Ingestion points: SKILL.md, verdict-agent.md). While boundary markers and explicit sanitization are absent, the capability inventory (Bash, Read, WebFetch) is used within a rigid analytical framework that requires specific evidence evaluation, characterizing this as a standard operational surface for a diagnostic tool.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 07:36 AM