mint-ai 0.0.8

Minimalist AI translation CLI powered by LLMs

🌐 Other languages

🌿 Mint

Minimalist AI Translation CLI — Simple. Fast. Intuitive.

Mint is a single-binary, LLM-powered translation CLI. Set two environment variables and translate anything from the command line — files, piped output, or inline text. Built-in language detection, grammar correction, streaming output, and multi-language rotation.

export MINT_PROVIDER=google-genai
export MINT_API_KEY=your_key

mint -t ja "Good morning"         # おはようございます
echo "早安" | mint -t en          # Good morning
cat document.txt | mint -t fr     # translate a whole file

---

✨ Why Mint?

• Zero-config — Single binary; API keys via env vars, no config file pollution
• Multi-provider — Google Gemini, OpenAI, Anthropic, or local Ollama / LM Studio
• Smart detection — Auto-detects language on every call; language-neutral content (numbers, symbols) passes through unchanged
• Smart correction — Same-language input? Auto-corrects grammar & spelling instead of translating
• Streaming — Output streams in real-time, no waiting for long translations
• Composable — Pipe-friendly stdin/stdout; pairs seamlessly with grep, sed, xargs, and friends
• Secure — Untrusted input is isolated from model instructions via system/user message separation and per-request random-nonce delimiters; translating adversarial content cannot hijack the LLM's behavior

---

📋 Installation

Automated install (recommended)

macOS / Linux

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/min0625/mint/main/script/install.sh)"

Auto-detects OS and architecture (Linux/macOS, x86_64/arm64), installs to ~/.local/bin. Override with MINT_INSTALL_DIR or pin a version with MINT_VERSION=v1.0.0.

Windows (PowerShell)

irm https://raw.githubusercontent.com/min0625/mint/main/script/install.ps1 | iex

Auto-detects architecture (x86_64/arm64) and installs to $HOME\.local\bin. Override with $env:MINT_INSTALL_DIR or pin a version with $env:MINT_VERSION = 'v1.0.0'.

Homebrew (macOS / Linux)

brew install min0625/tap/mint-ai

pipx

pipx install mint-ai

npm

npm install -g mint-ai

Manual download

Download the pre-built binary for your platform from GitHub Releases, move it into a directory on your PATH, then verify:

mint --version

---

🚀 Quick Start

1. Set your provider

# Google Gemini (free tier available — https://aistudio.google.com/apikey)
export MINT_PROVIDER=google-genai
export MINT_API_KEY=your_gemini_api_key

# OpenAI
export MINT_PROVIDER=openai
export MINT_API_KEY=sk-...

# Anthropic
export MINT_PROVIDER=anthropic
export MINT_API_KEY=sk-ant-...

# Ollama (no API key needed)
export MINT_PROVIDER=openai
export MINT_BASE_URL=http://localhost:11434
export MINT_MODEL_NAME=qwen2.5:7b  # use any model loaded in Ollama

# LM Studio (no API key needed)
export MINT_PROVIDER=openai
export MINT_BASE_URL=http://localhost:1234
export MINT_MODEL_NAME=lmstudio-community/Qwen2.5-7B-Instruct-GGUF  # use any model loaded in LM Studio

2. Translate

mint --target ja "Good morning"
mint -t zh-TW "Good morning"

echo "The quick brown fox" | mint -t fr
cat document.txt | mint -t zh-TW

Use --verbose / -v (or MINT_VERBOSE=true) to print diagnostic info and token usage to stderr:

mint -t ja -v "Good morning"
# [mint] provider: google-genai
# [mint] model: gemini-3.1-flash-lite
# [mint] single target — skipping language detection
# [mint] target language: ja
# おはようございます
# [mint] tokens: 113 in / 2 out

Typical token usage (measured on gemini-3.1-flash-lite):

Mode	Input	Calls	Input tokens	Output tokens
Single-target (-t or single MINT_TARGET_LANG)	short word/sentence	1	~110–130	~1–15
Single-target	long article (testdata/sample.txt)	1	~465–470	~450–560
Multi-target rotation (comma-separated MINT_TARGET_LANG)	short sentence	2	~250–260	~2–8
Explicit source -s + rotation	short sentence	1	~105–120	~1–2

Token counts scale with input length. Output tokens vary by target language — Japanese and Chinese tend to produce more tokens than English for equivalent content.

How far does 1M tokens go? (input + output combined, derived from the measured usage above):

Input	~Tokens per translation	Translations per 1M tokens
Short word or phrase	~120	~8,000
300-word article	~1,000	~1,000

Counts combine input and output tokens. Providers price input and output separately and many offer free tiers — check your provider's pricing page for current rates. Google Gemini's free tier at Google AI Studio needs no credit card.

Force the source language with --source / -s to translate input that is also valid in the target language (cross-language homographs, romanized text):

mint -s fr -t en "pain"          # French → bread (without -s, treated as English "pain")
mint -s ja -t en "konnichiwa"    # romaji Japanese → hello

3. Smart language detection

Translation with auto-detection:

export MINT_TARGET_LANG=en

mint "早安"   # Detects Chinese → Good morning

Grammar & spelling correction — when input language matches the target, Mint corrects instead of translates:

export MINT_TARGET_LANG=en

mint "Good mooorning"          # Detects English → Good morning
mint "She don't know nothing"  # Detects English → She doesn't know anything
mint "i luv coding"            # Detects English → I love coding

Language rotation — translates to the next language in the list, wrapping around:

# Two languages
export MINT_TARGET_LANG=en,zh-TW
mint "Hello"   # en → zh-TW: 你好
mint "你好"    # zh-TW → en: Hello

# Three languages
export MINT_TARGET_LANG=en,zh-TW,ja
mint "Hello"       # en → zh-TW: 你好
mint "你好"        # zh-TW → ja: こんにちは
mint "こんにちは"   # ja → en: Hello

---

🔑 Environment Variables

Variable	Description	Default
MINT_PROVIDER	google-genai | openai | anthropic	— (required)
MINT_API_KEY	API key; required when using the default endpoint; optional when MINT_BASE_URL is set (proxy handles auth)	—
MINT_BASE_URL	Custom API base URL (domain only; each provider appends its own path); use with openai to target Ollama (http://localhost:11434), LM Studio (http://localhost:1234), or any other OpenAI-compatible endpoint	Provider default
MINT_MODEL_NAME	Model to use; required when MINT_BASE_URL is set	gemini-3.1-flash-lite / gpt-4o-mini / claude-haiku-4-5
MINT_TARGET_LANG	Target language(s), e.g. en or en,zh-TW,ja	System locale
MINT_VERBOSE	Set to true to enable verbose diagnostic output (equivalent to --verbose)	false

---

🚩 CLI Flags

Flag	Short	Description
--target <lang>	-t	Target language (BCP-47 tag, e.g. ja, zh-TW, fr). Overrides MINT_TARGET_LANG.
--source <lang>	-s	Source language (BCP-47 tag); skips auto-detection and forces translation from this language.
--verbose	-v	Print diagnostic info and token usage to stderr. Also enabled by MINT_VERBOSE=true.
--version		Print version and exit.

---

📅 Roadmap

• Multi-LLM provider support (Google Gemini, OpenAI, Anthropic, local via Ollama / LM Studio)
• Smart language detection and multi-language rotation via MINT_TARGET_LANG
• Explicit target language via --target / -t flag
• Explicit source language via --source / -s flag
• Streaming output
• GoReleaser multi-platform binary release (Linux / macOS / Windows)
• Batch translation mode
• Glossary / custom dictionary support
• Output format options (plain text, JSON, Markdown)
• Caching for repeated translations

---

📄 License

Apache License 2.0 — see LICENSE for details.