Sofagent FDE Enterprise AI Governance Diagnostics

Problem It Solves

Enterprise AI governance diagnostics often break down in three concrete places: business context is scattered across documents and chats; AI handoff boundaries are decided by gut feel without auditable evidence; and the diagnostic output rarely becomes a reusable governance asset. FDE Skill targets this workflow by splitting the engagement into entry filing, ontology discovery, quantification, delivery, and exit, and by requiring the agent to load the stage-specific skill before acting.

How It Works And Where It Fits

It injects identity and baseline rules first, then reads think.md, enterprise norms, and a knowledge index as layered constraints. During execution, the agent inventories process elements such as input, output, owner, time cost, and pain point, builds entity/concept/relations ontology data, and performs three-question classification plus ROI estimation. With the MCP Server connected, it can use tools like run_audit, get_think, fde_quantify, and snapshot_restore; without it, the flow degrades to text-guided operation. Audit results keep the [sofagent] prefix for traceability. It fits teams diagnosing business processes, evaluating AI takeover boundaries, and distilling a company-specific skill. It is not a substitute for writing business code, bypassing human review, or handling sensitive data without masking.