Run Analysis

Stage 1: Geometry → Stage 2: Sections → Stage 3: Materials & Analysis

Idle Ready to run.

🔑 Agent LLM Settings

Use defaults (DEEPSEEK_MODEL)
Structure Type:
Modeling Fidelity:
Output Platform:
0 characters ~0 tokens

Pipeline Progress

0/6 agents ready

2. Sections & Material Properties

Probabilistic characterization of material properties · constitutive model details

2 🔲 Sections — refined cross-section assignment & review
🔲 Section Geometry — fiber cross-sections from the generated model · click for detail

Run the pipeline to see cross-section geometry.

3 🔬 Materials & Analysis — material uncertainty · MC simulation (downstream)

🔬 Material Uncertainty

Run the pipeline to extract material uncertainty information from your prompt.

Uncertainty Parameters

No material uncertainty data yet.

Constitutive Model Details

Constitutive model information will appear here after pipeline execution.

3. MC Simulation

Downstream: parallel sampling of material uncertainty (after geometry & sections are defined)

Simulation Control

MC settings derived from the pipeline prompt. Run the pipeline first.

1. Geometry — Structural Visualization

Detailed node/element inspection, geometry validation & consistency checks

🖱 Hover over elements to see section dimensions and material properties. Click to flag issues.

🏗️ Overall Frame Geometry — structural grid overview (grid intersections only)
No geometry data yet. Run the pipeline to see the structural frame.
Column Beam
🔍 Detailed Frame Geometry — drag to pan · scroll to zoom · rotate · hover elements for section info
No node / element data yet.
| Rotate |
Node Column Beam 🖱 hover: section + material · click: flag issue
✅ Geometry Consistency Check

Run the pipeline to see automatic geometry validation results.

🔍 Error Feedback

0 item(s) selected

Click nodes/elements in the diagrams above to select them, then classify each error type below.

No items selected. Click a node or element in the visualization to begin.

Benchmark Batch Runner

Run randomized test cases and review results

Idle Ready to run benchmark cases.

Prompt Run History

Recent submitted prompts and their outcomes

RL Optimization & Feedback

Reward signals and human feedback collection

Reward Score

0.00Total Reward
none
Error Type
observe
Policy Action
SQLite
Database

Reward Components

Human Feedback

Mark the current result

Mark the current result as correct or incorrect. This feedback is incorporated into RL reward computation.

Prompt Lifecycle

Production

Run stages in order: Optimization → Confirmation → Production. Production uses only promoted frozen prompts.

Loading candidate prompts...

🧬 MIPROv2 Prompt Optimizer

Idle

Generates candidate prompts from recorded successes and failures during Optimization. Candidates must still pass Confirmation before they can be promoted to Production.

Run optimization to see results here. Each round uses the LLM to analyze failure cases and propose improved prompts.

Code Execution

Run generated code and view section force diagrams

Idle Ready to execute generated code.

Stdout

Stderr

Complete Code

Generated OpenSeesPy code

        

Python Check Agent Diagnosis


        

Section Force Diagrams

Idle

Run generated code to render diagrams.

Axial Force (N)
Axial force diagram
Not generated yet.
Shear Force (V)
Shear force diagram
Not generated yet.
Bending Moment (M)
Bending moment diagram
Not generated yet.

Agent Rewards

Per-agent RL reward decomposition and experience summary

Loading agent reward data...