Fuel Terminal Logistics Exception Handling with Claude Agents SDK & Aider
A Turkish company signed a contract for a major fuel storage terminal in Libya. Build an autonomous exception-handling agent using Claude Agents SDK and Aider to analyze shipment delay telemetry, automatically patch dispatch scripts, and re-route fuel tankers around port congestion. Achieve automated resolution on over 80% of supply chain delays.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop a Claude Agents SDK workflow enhanced by Aider for code generation to automatically handle supply chain anomalies and patch routing scripts.
How work is evaluated
Tests agent ability to inspect log exceptions and run script patches for tanker rerouting.
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.
valid_alternate_port
Ensures alternate port accepts specific fuel type
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
delay_reduction_ratio
Ratio of hours saved over total original congestion delay • target: 0.75 • 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
Implement extended thinking patterns using Claude Agents SDK to reason about port bottlenecks.
Leverage Aider CLI mechanisms to programmatically update Python dispatch scripts.
Process vessel AIS telemetry streams and fuel storage inventory levels.
Formulate automated rerouting logic with minimal fuel transport cost markup.
Reference links and supporting material
Terminal status logs and tanker position data across Mediterranean and North African fuel corridors.
Challenge setup
Tests agent ability to inspect log exceptions and run script patches for tanker rerouting.
Challenge input
JSON alert object with port status and vessel queues
Claude Agents SDK
Provides strong reasoning and multi-tool orchestration.
Aider
Enables dynamic code editing and script patching.
Evaluated output
JSON response with modified tanker route and generated script diff
- Ensures alternate port accepts specific fuel type
- Ratio of hours saved over total original congestion delay • target: 0.75 • range: 0-1
- Benchmark: LogisticsAgentBench
- Delay Reduction Ratio target: 0.75
- 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
- Aider · Required
- Bland AI · Optional
- Agency (Observability) · Optional
- Aider · Required
Evaluation contract
- valid_alternate_port · Weight 1
- delay_reduction_ratio · Weight 1
Metrics
- delay_reduction_ratio · Target 0.75
Benchmarks
- LogisticsAgentBench
- security
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON alert object with port status and vessel queues
Output: JSON response with modified tanker route and generated script diff
Execution harness: Python
Evaluation tools
- Claude Agents SDK · Agent System
- Aider · AI Engineering Tooling
Recipe freeze marker
No recipe freeze time is recorded.