AV
In this project, I used gemini argon to create a PCB simulator and detector that examined a PCB section at a certain snapshot and evaluated any critical design flaws, specifically from a benchmark of 6 categories.
Core trick
Argon scores PCB section snapshots against six failure categories, ranks candidate flaws, and highlights them in a 3D render.
Description
In this project, I used gemini argon to create a PCB simulator and detector that examined a PCB section at a certain snapshot and evaluated any critical design flaws, specifically from a benchmark of 6 categories. It used argon 4 to do a verification loop to rank the possibility of the critical point being one of the 6 failures, and output a finalized 3D simulation with highlighted critical flaws it detected.
alvolkov07/Gemini-Argon-4-SF-Hackathon★ 00 forksJavaScriptMITCreated Oct 25, 2026 · last push Oct 25, 2026
README excerpt
# Gemini 4 Argon: AI PCB Inspector & Design Rule Checker An a PCB design verification estimator agent powered by **Gemini 4 Argon** (`gemini-4-argon`) via the **Interactions API**. The agent analyzes PCB layout diagrams, schematics, and Gerber preview images to detect potential electrical, manufacturing, and signal integrity flaws, classifying them by severity and recommending actionable remediation. The data was tested and evaluated from an existing dataset of zoomed-in snapshots of PCB sections: https://github.com/tangsanli5201/DeepPCB/tree/master/PCBData. --- ## Features - **Gemini 4 Argon Reasoning (`thinking_level="high"`):** Harnesses deep multi-step reasoning to evaluate high-speed traces, loop inductances, acid traps, and thermal dissipation paths. - **Structured Flaw Classification:** Categorizes defects into DRC violations, Signal Integrity (SI), Power Integrity (PI), Thermal Management, and DFM/Assembly yield risks. - **Automated Synthetic Demo Generator:** Generates a synthetic test PCB with intentional flaws (e.g., acute angle acid traps, misplaced decoupling capacitors, solder bridge clearances) to test without external data. - **Formatted Reports:** Generates Markdown and machine-readable JSON reports with an overall design health score. --- ## Quickstart ### 1. Installation Clone the repository and install the dependencies: ```bash python3 -m pip install -r requirements.txt ``` ### 2. Configure API Key Create a `.env` file in the project root: ```