Openiclib

Open Integrated Circuit Library — a curated catalog of silicon-proven IC designs from open PDKs.

---

Table of Contents

• Overview
• Features
• Supported PDKs
• Getting Started
• Usage
  • Browse the Library
  • CLI
  • Discovery Pipeline
• Architecture
• Contributing
• Citing
• License

---

Overview

Openiclib aggregates taped-out, silicon-proven IC designs built on open-source process design kits (PDKs). It provides a searchable, filterable catalog that researchers, startups, and students can use as a reference library for IC system-level design.

The project automatically discovers design repositories on GitHub and GitLab, classifies them by PDK, circuit class, and circuit type using a keyword + TF-IDF classifier, and publishes the results as a browsable website with interactive statistics.

Openiclib is affiliated with gdsfactory.

Features

• Automated discovery — crawls GitHub and GitLab for IC design repos using topic search, keyword search, org crawling, and tapeout repo subdirectory enumeration
• Keyword + TF-IDF classification — classifies designs by PDK, circuit class (analog/digital/mixed-signal/optical), circuit type, and silicon-proven status using scikit-learn (no API keys needed)
• Tapeout repo crawling — enumerates individual designs inside IHP-GmbH tapeout repos and TinyTapeout shuttle repos
• Filterable web catalog — MkDocs-powered site with GitHub-style cards, search, dropdown filters, and silicon-proven toggle
• Interactive statistics — ECharts-based dashboards showing design distributions by PDK, circuit class, type, and fabrication status
• CLI tools — list, validate, discover, and classify designs from the command line
• Weekly CI pipeline — GitHub Actions workflow that discovers, classifies, and opens a PR with database updates

Supported PDKs

PDK	Foundry	Node
SkyWater SKY130	SkyWater Technology	130 nm CMOS
IHP SG13G2	IHP Microelectronics	130 nm BiCMOS
GlobalFoundries GF180MCU	GlobalFoundries	180 nm CMOS

Getting Started

Prerequisites

• Python 3.12+
• uv package manager

Installation

git clone https://github.com/gdsfactory/openiclib.git
cd openiclib
uv sync

Quick Start

# List all designs in the database
uv run openiclib db list

# Filter by PDK
uv run openiclib db list --pdk ihp130sg

# Filter silicon-proven designs only
uv run openiclib db list --proven

# Validate the database
uv run openiclib db validate

Usage

Browse the Library

Visit the live site at openiclib.github.io to browse, search, and filter designs interactively.

To serve the docs locally:

uv run mkdocs serve

CLI

# List designs with filters
uv run openiclib db list --pdk skywater130 --circuit-class analog

# Validate database integrity
uv run openiclib db validate

# Discover candidate repos (dry run)
uv run openiclib discover --source github --dry-run

# Classify candidates into the design database
uv run openiclib classify --input candidates.json --output data/designs.json -v

Discovery Pipeline

Run the full pipeline to discover, classify, and validate:

make pipeline

This executes:

1. Discover — searches GitHub/GitLab for IC design repos and crawls IHP + TinyTapeout tapeout directories
2. Classify — runs the keyword + TF-IDF classifier on all candidates
3. Validate — checks database integrity

Set GITHUB_TOKEN for higher API rate limits:

GITHUB_TOKEN=ghp_... make pipeline

Architecture

GitHub/GitLab APIs ──→ discover.py ──→ candidates.json
                                           │
                            keyword_classifier.py (TF-IDF + scikit-learn)
                                           │
                                      designs.json
                                           │
                            catalog_hook.py (MkDocs on_pre_build)
                             │          │           │
                        library.md   stats.md   designs.json
                             │
                        MkDocs build ──→ GitHub Pages

Key Modules

Module	Purpose
models.py	Pydantic data models (Design, DesignDatabase, enums)
db.py	Load/save/validate JSON database + protobuf conversion
discover.py	GitHub discovery (topics, keywords, orgs, tapeout repos)
discover_gitlab.py	GitLab discovery
keyword_classifier.py	TF-IDF + keyword classification (scikit-learn)
cli.py	Click CLI (db list, db validate, discover, classify)
stats.py	Statistics generation
catalog_hook.py	MkDocs hook generating library and stats pages

Design Database

The canonical database is data/designs.json. Each design is classified by:

• PDK: skywater130, ihp130sg, gf180, gf90, gf45
• Circuit class: analog, digital, mixed-signal, optical
• Circuit type: LNA, ADC, PLL, RISC-V Core, SoC, etc.
• Silicon proven: fabricated and measured

The protobuf schema in data/designs.proto serves as the formal schema definition.

Contributing

Contributions are welcome! You can:

1. Submit a design — open a PR adding your design to data/designs.json
2. Improve classification — add keywords to keyword_classifier.py for better accuracy
3. Add discovery sources — extend discover.py with new orgs, topics, or tapeout repos
4. Report issues — open an issue for bugs or feature requests

Development

uv sync                      # Install dependencies
uv run pytest                # Run tests (116 tests)
uv run ruff check .          # Lint
uv run mkdocs build --strict # Build docs

Citing

If you use Openiclib in your research, please cite:

@software{dias2026openiclib,
  author       = {Dias, Diogo},
  title        = {Openiclib: Open Integrated Circuit Library},
  year         = {2026},
  url          = {https://github.com/das-dias/openiclib},
  license      = {MIT},
  note         = {A searchable catalog of silicon-proven IC designs from open PDKs (SKY130, IHP SG13G2, GF180)}
}

License

This project is licensed under the MIT License. See LICENSE for details.