Automated Crypto-Fiat Banking Gateway Guard with OpenAI Agents SDK
Banking crypto clients requires rigorous transaction provenance tracking and automated Travel Rule compliance validation before fiat conversion. Build an autonomous gateway guard using the OpenAI Agents SDK to analyze blockchain wallet histories, compute risk scores, and enforce automated threshold controls. Maintain sub-second processing speed with 98% compliance accuracy on high-value transfers.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop an autonomous crypto-fiat settlement gateway guard using OpenAI Agents SDK to verify on-chain transaction provenance and Travel Rule compliance.
How work is evaluated
Evaluates accuracy of Travel Rule verification and crypto risk scoring decisions.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
travel_rule_enforcement_test
Checks if transfers > $3,000 missing Travel Rule info are QUARANTINED
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
compliance_precision
Precision score of compliance clearance decisions • target: 0.98 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Initialize OpenAI Agents SDK agents with custom tool function definitions
Perform automated verification of counterparty VASP identity and Travel Rule metadata
Evaluate wallet cluster risk using mock blockchain explorer tools
Automate fiat minting/burning clearance based on risk thresholds
Reference links and supporting material
Dataset of 300 crypto transfer records with varying compliance completeness, mixing missing VASP data with legitimate transfers.
Challenge setup
Evaluates accuracy of Travel Rule verification and crypto risk scoring decisions.
Challenge input
JSON with source_address, target_bank_acc, amount_usd, and origin_vasp
OpenAI Agents SDK
Official OpenAI Agent orchestration
Evaluated output
JSON with disposition (APPROVED, QUARANTINED, REJECTED), travel_rule_valid, and risk_score
- Checks if transfers > $3,000 missing Travel Rule info are QUARANTINED
- Precision score of compliance clearance decisions • target: 0.98 • range: 0-1
- Benchmark: CryptoGuard-Eval
- Compliance Precision target: 0.98
- 1 public reference case
- 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
- OpenAI · Required
- Azure OpenAI Service · Optional
- OpenAI o3 · Optional
Evaluation contract
- travel_rule_enforcement_test · Weight 1
- compliance_precision · Weight 1
Metrics
- compliance_precision · Target 0.98
Benchmarks
- CryptoGuard-Eval
- security
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with source_address, target_bank_acc, amount_usd, and origin_vasp
Output: JSON with disposition (APPROVED, QUARANTINED, REJECTED), travel_rule_valid, and risk_score
Execution harness: Python
Evaluation tools
- OpenAI Agents SDK · Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.