Corporate Credit Due Diligence

Problem It Addresses

Bank corporate credit teams often validate customers across disconnected systems: business registration, equity structure, financial statements, operating data, group relations, and compliance risks. The skill targets corporate lending, credit pre-diligence, and risk-control workflows, turning initial screening, corporate profiling, pre-diligence, financial review, group analysis, and customer discovery into a repeatable process. It emphasizes evidence-backed conclusions with clickable source links.

How It Works

• Dynamic buddy selection: reads the live GET /api/playbook response and uses current buddies[] instead of a hardcoded template list.
• Semantic matching: delegates to cue-research +match/Stage-2 to match the user's subject and goal, and asks for confirmation when hits are weak.
• Credit confirmation: requires explicit user approval for the subject, selected buddy, and credits consumption before running deep research.
• Execution: runs via research_run.py, reads the final RESULT, and must not fabricate output when the result is empty.
• Report delivery: covers corporate profiling, credit pre-diligence, group strategy analysis, high-net-worth customer discovery, financial due diligence, and initial screening memos.

Boundaries and Cautions

It only covers public data and does not replace legal, compliance, underwriting, or full due-diligence judgments. Private data, internal supervisory data, or unsupported non-public conclusions are out of scope. Runtime depends on a Cue account, git, and python3; network timeouts, insufficient credits, or missing live report segments should be handled by the documented prompts or replay path.