jazz1x/honne

README preview

# honne

> Claude Code plugin — self-observation from LLM transcripts

![version](https://img.shields.io/badge/version-0.0.3-blue)
![license](https://img.shields.io/badge/license-MIT-green)
![claude-code](https://img.shields.io/badge/claude--code-plugin-purple)

**honne** (本音, "true voice") — a local, evidence-backed mirror of how you actually work with LLMs. Beneath the official persona (*tatemae*), honne surfaces what your transcripts quietly record: recurring vocabulary, rejected suggestions, session rituals, the patterns you never named.

Everything runs locally. No network calls, no analytics. Claims are extracted and recorded autonomously; narrative explanations are synthesized via LLM in a single dedicated step. Your data lives under `.honne/` as plain JSONL — portable, inspectable, deletable.

[한국어](./README.ko.md) · [日本語](./README.jp.md)

## Skills

| Skill | Command | Role |
|-------|---------|------|
| **whoami** | `/honne:whoami` | Main orchestrator. 7-axis persona with autonomous evidence gathering and LLM narrative synthesis. |
| **lexi** | `/honne:lexi` | Lexicon axis only (high-frequency vocabulary, code-switching, onomatopoeia). |

View full repository on GitHub →