I built an open-source AI agent for EasyEDA that can design schematics and PCBs

Description and use

EasyEDA Copilot automates schematic and PCB design using an AI agent, MCP middleware, and placement and routing algorithms, while producing standard EasyEDA documents.

Problem

Designing schematics and PCBs manually in EasyEDA is time‑consuming and error‑prone.

Outcome

EasyEDA Copilot reduces manual effort by automating schematic creation, component placement, and PCB routing, while still allowing manual intervention via JavaScript API.

Workflow

• 1. AI agent generates a DSL description of board geometry, component groups, and placement constraints.
• 2. MCP server connects to EasyEDA via local WebSocket, exposing schematic and PCB data to the agent.
• 3. Placement algorithm receives DSL, board geometry, and constraints, then computes optimal component positions using GPU acceleration.
• 4. Agent reviews placement preview, adjusts constraints if needed, and applies placement to EasyEDA.
• 5. Routing algorithm receives DSL describing trace widths, clearances, differential pairs, and planes, then runs KiCad Routing Tools via an integration.
• 6. Agent imports routed traces back into EasyEDA, checks for remaining unrouted connections, and runs EasyEDA’s built‑in DRC.
• 7. Checkpoint system saves document state before changes, allowing rollback if routing fails.

Evidence

EasyEDA Copilot can handle a large part of the design process, and the result is still a normal, editable EasyEDA document.

Source

The MCP server connects to the EasyEDA extension through a local WebSocket.

Source

Placement uses a separate algorithm that receives components, connections, board geometry, and constraints.

Source

The current implementation uses KiCad Routing Tools.

Source