KiCad AI Skill: An Open-Source Intelligent Toolchain for Automating EDA Design Flows

Core Introduction

KiCad AI Skill is an open-source AI skill project based on the Agent Skills standard. It enables automated analysis, verification, repair, and manufacturing export of KiCad projects via the kicad-tools CLI, compatible with mainstream AI tools like GitHub Copilot and Google Antigravity. It addresses the pain points of extensive manual operations in traditional hardware design workflows, marking a new stage in hardware design automation.

Background Overview

Electronic Design Automation (EDA) is a core part of hardware development. KiCad, as an open-source EDA software, is used by millions of engineers worldwide. However, traditional workflows rely heavily on manual operations: from schematic checking to PCB layout verification, from DRC/ERC checks to manufacturing file export—every step requires careful review by engineers. The michpro/kicad-tools project integrates AI capabilities into the KiCad workflow, allowing AI tools to directly understand and manipulate KiCad files through the open Agent Skills standard.

Key Features

1. Schematic Analysis: Symbol list, net tracing, hierarchy verification, BOM generation
2. PCB Analysis: Board-level summary, footprint tracing, net analysis, stackup structure
3. DRC/ERC Verification: Pure Python DRC, manufacturer rule checks (JLCPCB, etc.), ERC checks
4. Automatic Repair: Schematic (symbol replacement, parameter setting, etc.), PCB (spacing repair, DRC correction, etc.)
5. Automatic Routing: Supports multi-layer boards, checkpoint mechanism, automatic DRC repair
6. Layout Optimization: CMA-ES optimizer, anchor weight locking for key components
7. Manufacturing Export: Gerber/BOM/CPL generation, PDF report, production readiness verification
8. Library Tools: Footprint generation, LCSC search, datasheet parsing

Architecture & Compatibility

• Agent Skills Standard: Adopts open standards to ensure interoperability, supported tools include GitHub Copilot (native), Google Antigravity (native), Claude Code (native), Cursor (needs adaptation)
• Project Structure: kicad-tools/ contains SKILL.md (main instructions), cli-reference.md (50+ commands), workflows.md (6 multi-stage workflows)
• Workflows: Schematic review/repair, PCB review/repair, manufacturing export, complete build pipeline

Interaction Flow

Follows the principle of progressive disclosure:

1. Discovery: AI sees the skill name and description
2. Activation: Loads SKILL.md when the user mentions KiCad/EDA tasks
3. Guidance: Checks kct version; if missing, automatically creates a virtual environment and installs it
4. Execution: Runs kct commands (--format json) and parses the results
5. Progressive Disclosure: For complex tasks, refers to cli-reference or workflows documents

Installation Steps

• GitHub Copilot (VS Code): Copy kicad-tools to the .github/skills/ directory
• Google Antigravity: Copy to the .agents/skills/ directory
• Automatic Installation: When kicad-tools is not installed, AI automatically creates a virtual environment, installs from PyPI, and verifies

Significance & Outlook

• Milestone Value: Standardized interoperability, no need for MCP servers, progressive automation, open-source ecosystem promotes democratization of hardware design
• Future Directions: Further intelligence (automatic layout optimization, intelligent signal integrity analysis, component selection recommendations, etc.), laying the foundation for AI-assisted hardware design