Skip to content
DEVELOPERS · AGENT SDK

Built for IC design
Agent developers

Call from Python, integrate via MCP, or orchestrate with Tcl—iEDA.ai keeps one API experience. In five minutes, let your LLM drive the first placement.

5-minute quickstart → API reference → GitHub ↗
Agent SDK
# One API, three protocols
Python   │ client.call("iPL.place", ...)
MCP      │ {"tool": "iPL.place", ...}
Tcl      │ iPL::place -effort standard
# Same iDB, same result
iMapsynthesisiFPfloorplaniPDNpoweriPLplacementiCTSclockiTOoptimizationiRTroutingiSTAtimingAiEDAdesign dataiPCLlayout modeliMapsynthesisiFPfloorplaniPDNpoweriPLplacementiCTSclockiTOoptimizationiRTroutingiSTAtimingAiEDAdesign dataiPCLlayout model
CODE

Three protocols, one experience

Whichever call style you prefer, iDB and the toolchain stay the same.

from ieda import AgentClient

client = AgentClient()
design = client.load_design("aes_core", tech="sky130")
result = client.call("iPL.place", design=design, effort="standard")

print(f"HPWL: {result.hpwl}, Density: {result.density}")
JOURNEY

Developer journey

From quickstart to deep integration—docs and examples at every step.

01

Quickstart

Run your first tool call in five minutes—load design, place, timing analysis.

Get started 02

API reference

Full Python/MCP/Tcl API docs with every endpoint and parameter documented.

API docs 03

SDK guide

SDK installation, configuration, error handling, and best practices—from initialization to production deployment.

SDK guide 04

Contract spec

Typed request/result/proposal/delta contracts—the Agent–tool agreement.

Contract docs 05

Integration examples

End-to-end Agent scenarios: timing repair, placement optimization—runnable examples.

View examples 06

GitHub

Open-source repo—star, contribute, open issues and PRs, build Agent-native EDA together.

GitHub

Ready to build your first
IC design Agent?

Start with Quickstart—drive chip placement with an LLM in five minutes.

5-minute quickstart → SDK guide