Skip to main content
Back to challenges
Challenge

Build MSME Credit Underwriter with Claude Agents SDK and o3 Reasoning

MSME credit decisioning requires accurate risk assessment across non-standardized bank statements and tax filings. Build an autonomous underwriting agent using Claude Agents SDK and o3 deep reasoning models to calculate risk scores and credit limits with complete regulatory transparency.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an MSME credit assessment engine combining Claude Agents SDK tools and o3 model reasoning for Account Aggregator financial records.

Delivery guide

How work is evaluated

Evaluation

Evaluates underwriting decision accuracy and transparency against benchmark MSME loan histories.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

audit_trail_exists

Checks that complete audit trail reasoning was generated

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

underwriting_accuracy

Accuracy relative to gold standard credit outcomes • target: 0.92 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Construct multi-turn financial document reasoning workflows with Claude Agents SDK

  • Leverage OpenAI o3 model reasoning capabilities for complex cash flow analysis

  • Generate explainable credit score outputs aligned with RBI lending standards

Resources and assets

Reference links and supporting material

Dataset notes

Anonymized MSME financial profiles and Account Aggregator transaction summaries.

Challenge setup

Evaluates underwriting decision accuracy and transparency against benchmark MSME loan histories.

Challenge input

JSON containing annual turnover, debt service ratio, and GST reconciliation delta

Claude Agents SDK

Provides agent orchestration and robust tool calling

o3

Delivers deep reasoning for financial auditing

Evaluated output

JSON containing decision, sanctioned limit, credit score, and reasoning breakdown

Evaluation checks
  • Checks that complete audit trail reasoning was generated
  • Accuracy relative to gold standard credit outcomes • target: 0.92 • range: 0-1
Configured evaluation evidence
  • Benchmark: MSME-Credit-Bench-v2
  • Underwriting Accuracy target: 0.92
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Writer · Optional
  • Zed · Optional
Policy Serving
  • o3 · Required

Evaluation contract

  • audit_trail_exists · Weight 1
  • underwriting_accuracy · Weight 1

Metrics

  • underwriting_accuracy · Target 0.92

Benchmarks

  • MSME-Credit-Bench-v2
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing annual turnover, debt service ratio, and GST reconciliation delta

    Output: JSON containing decision, sanctioned limit, credit score, and reasoning breakdown

    Execution harness: Python

Evaluation tools

  • Claude Agents SDK · Agent Framework
  • o3 · Reasoning Model

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build MSME Credit Underwriter with Claude Agents SDK and o3 Reasoning