Business diagnosis
Ability to interview owners and users, and break down triggers, steps, inputs, outputs, exceptions, value, and stop conditions.

FDE Capability Guide
It doesn't matter how many hours of courses you've watched; what matters is whether you can provide evidence of tasks, systems, evaluations, launches, and handovers in real business.
Each dimension should correspond to real artifacts. Being able to explain concepts does not equal being able to take on on-site delivery.
Ability to interview owners and users, and break down triggers, steps, inputs, outputs, exceptions, value, and stop conditions.
Can trace sources, fields, permissions, interfaces, update frequency, conflicts, failures, and existing technical debt.
Can design models, retrieval, tool calls, state, human confirmation, and fallback - and turn them into a runnable version.
Can establish representative samples, quality and permission checks, rollout criteria, audit, and human disposition.
Can handle deployment, monitoring, cost, performance, alerts, knowledge updates, versioning, and support.
Can manage scope, communication risk, push for confirmation, document evidence, train users, and complete internal handover.
Levels are used to choose the next real task, not as a permanent identity. Re-assess whenever the system and scope of responsibility change.
Can restate business processes, identify data and system boundaries, and complete structured records under mentor guidance.
Can build a bounded workflow, tool connections, and representative tests with pair support.
Can complete permissions, evaluation, monitoring, rollback, and human takeover, and go into production within a limited scope.
Can independently lead business interviews, technical decisions, launch retrospectives, and knowledge transfer, and mentor the next engineer.
Choose a real business line you recently worked on, bring all materials, then decide the next stage.
Problem map, flow diagram, responsible person, and clear success and stop conditions.
System map, code or configuration, interface and permission records, versions, and failure handling.
Normal and abnormal samples, evaluation results, human judgment, rollout criteria, and unresolved issues.
Release, monitoring, alerts, rollback, usage feedback, retrospectives, and internal handover records.

State the background of the person to be developed, the target business line, and the current system; we design the development starting point based on verifiable artifacts.