AI SDK Module Type Package Orchestrator with Replit Agent Integration
Module Type Package (MTP) standardization enables modular plug-and-produce automation, but dynamic state synchronization remains challenging. Build a TypeScript agent using Vercel AI SDK and Replit Agent workflow patterns to parse MTP manifest files and orchestrate unit execution states. Success is verified by achieving 100% deterministic MTP service state machine transitions without deadlocks.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop an AI SDK streaming agent configured via Replit Agent tooling to dynamically parse MTP XML/JSON manifests and trigger process state machine transitions.
How work is evaluated
Evaluates MTP manifest parsing and state transition command generation.
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_state_transitions
Ensures no prohibited jumps (e.g. IDLE directly to EXECUTE without STARTING).
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
transition_correctness
Accuracy of generated state sequence commands. • target: 1 • 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
Parse MTP Service and OPC UA NodeID definitions using AI SDK structured object generation.
Utilize Replit Agent patterns to dynamically assemble workflow execution code.
Implement state machine validation for MTP Idle, Starting, Execute, and Completing states.
Stream live state machine progress back to process control dashboards.
Reference links and supporting material
JSON/XML representations of Modular Type Package (MTP) manifests for chemical reactor dosing and heating units.
Challenge setup
Evaluates MTP manifest parsing and state transition command generation.
Challenge input
JSON object with mtp_manifest and target_command
AI SDK
Required framework for building streaming TypeScript agents.
Replit Agent
Utilized for automated workflow code synthesis and tool assembly.
Evaluated output
JSON object with state_sequence, commanded_nodes, and success_flag
- Ensures no prohibited jumps (e.g. IDLE directly to EXECUTE without STARTING).
- Accuracy of generated state sequence commands. • target: 1 • range: 0-1
- Benchmark: MTPAutomationBench-2026
- Transition Correctness target: 1
- 1 public reference case
- JavaScript execution harness
- javascript 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
- Replit Agent · Required
- Replit Agent · Required
- Vercel AI SDK · Optional
- Vercel AI SDK · Optional
Evaluation contract
- valid_state_transitions · Weight 1
- transition_correctness · Weight 1
Metrics
- transition_correctness · Target 1
Benchmarks
- MTPAutomationBench-2026
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object with mtp_manifest and target_command
Output: JSON object with state_sequence, commanded_nodes, and success_flag
Execution harness: JavaScript
Evaluation tools
- AI SDK · Agent Framework
- Replit Agent · AI Engineering Tooling
Recipe freeze marker
No recipe freeze time is recorded.