A field guide to Japanese NLP that happens to be a Claude plugin

What it does

This is a curated awesome-list tracking 878 GitHub repositories and 370 Hugging Face models and datasets for Japanese NLP — tokenizers, parsers, dictionaries, corpora, tutorials, and pretrained models across Python, C++, Rust, JavaScript, Go, and Java. The list is CC0-licensed and actively maintained, with download counts and commit freshness badges for popular Python libraries.

The interesting bit

The project ships a Claude Code plugin with four skills that query the bundled dataset. The search skill returns a “Use-case Selection Guide” table matching beginner scenarios to specific repos with popularity metrics. A find-new-resources skill trawls GitHub for repos not yet in the list — essentially automating its own maintenance. The plugin auto-detects Japanese queries and responds in Japanese.

Key highlights

• 1,200+ resources indexed across GitHub and Hugging Face
• Plugin skills cover search, trend research, issue investigation, and new-repo discovery
• Download and star statistics for major Python tokenizers (SudachiPy, Janome, fugashi, etc.)
• Multilingual READMEs: English, Japanese, Traditional Chinese, Simplified Chinese
• Companion survey site launched March 2026

Caveats

• Some listed tools show stale commits (MeCab forks, natto-py last updated November 2023)
• The Mecari entry has an invalid star count, suggesting incomplete curation
• Plugin installation requires Claude Code specifically; no standalone CLI or API

Verdict

Worth bookmarking if you work on Japanese NLP and want to stop reinventing the tool-discovery wheel. Skip it if you need a programmable API or work outside the Claude Code ecosystem — the raw data is open, but the query interface is tightly coupled.