mangopi-cli 0.1.18

Mango Pi Cli

Mangopi CLI

A lightweight, zero-dependency AI coding assistant running directly in your terminal.

Mangopi CLI is a local-first AI coding assistant inspired by tools like Claude Code, designed for developers who want a fast, hackable, and minimal runtime experience.

It provides an agentic coding workflow with:

• file editing
• shell execution
• tool calling
• autonomous goal execution
• context-aware conversation management
• automatic context compacting

All with instant startup and no heavyweight framework dependencies.

---

Features

• Zero dependency (Python standard library only)
• Instant startup speed
• Claude Code–style terminal UX
• OpenAI-compatible API support
• Built-in file and shell tools
• Autonomous Goal Mode
• Automatic context compacting
• Persistent local sessions
• Skill system support (SKILL.md)
• Safe shell execution checks
• Fully hackable and easy to extend
• Large-context optimized architecture

---

Installation

From PyPI

pip install mangopi-cli

Start Mangopi CLI:

mangopi-cli

---

From Source

git clone [e-mail]:w4n9H/mangopi-cli.git
cd mangopi-cli
python mangopi_cli.py

---

Configuration

Required:

export MANGO_KEY="your_api_key"

Recommended:

export MANGO_API_URL="https://api.deepseek.com"
export MANGO_MODEL="deepseek-v4-flash"

Optional:

export MANGO_MAX_CONTEXT=1000000
export MANGO_LANG=zh

---

Supported Providers

Mangopi CLI supports:

• DeepSeek
• OpenAI-compatible APIs
• MiniMax
• Custom compatible endpoints

Example:

export MANGO_API_URL="https://api.openai.com/v1"
export MANGO_MODEL="gpt-4o-mini"

---

Usage

Start the CLI:

mangopi-cli

or:

python mangopi_cli.py

---

Built-in Commands

Command	Description
/q	Quit
/n	Start a new session
/c	Compact current session
/h	Show help
/g <goal>	Autonomous goal execution mode

---

Goal Mode

Goal Mode allows Mangopi CLI to autonomously:

• plan
• execute
• verify
• iterate

until the objective is fully completed.

Example:

/g build a fastapi todo app with tests

The agent will continue working until it determines the task is complete.

---

Built-in Tools

Tool	Description
read	Read files
write	Write or overwrite files
edit	Replace exact strings in files
search	Search files using glob patterns
grep	Regex search through project files
bash	Execute shell commands
use_skill	Load installed skills
attempt_completion	Finish the current task

Mangopi CLI can autonomously inspect files, modify code, search projects, and execute shell commands.

---

Skill System

Mangopi CLI supports reusable workflow skills.

Example structure:

~/.mangocli/skills/python_backend/

├── SKILL.md
├── scripts/
└── references/

Example SKILL.md:

---
description: Python backend workflow
tags: ["python", "backend"]
---

Use pytest for tests.
Prefer small functions.

The model can automatically discover and load relevant skills during execution.

---

Session Persistence

Sessions are stored locally:

.mangocli/session/session.json

Mangopi CLI automatically:

• restores previous sessions
• preserves important context
• compacts old conversations
• manages long-running workflows

---

Context Compacting

Mangopi CLI includes multiple compacting strategies:

• micro compact
• session memory compact
• conversation compact
• full LLM summary compact

This enables extremely long-running coding sessions while staying within model context limits.

---

Safety

Dangerous shell commands require confirmation before execution.

Examples include:

• rm -rf
• mkfs
• chmod 777
• sudo rm
• destructive system operations

---

Architecture

Core components:

Component	Responsibility
Printer	Terminal UI rendering
ContextManager	Conversation memory & compacting
ToolBase	Tool framework
Provider	API abstraction layer
SystemPrompt	Runtime prompt assembly
SkillManager	Skill discovery & loading

---

Philosophy

Mangopi CLI focuses on:

• fast startup
• zero dependency
• local-first workflows
• terminal-native AI interaction
• lightweight runtime design
• simplicity over abstraction
• hackability over frameworks

No Electron. No Docker. No Redis. No heavyweight AI frameworks.

Just a fast and hackable AI coding assistant for the terminal.

---

License

Apache License 2.0

---

Author

Created by moofs.