core-termux

Turn Termux into a complete development workstation with AI coding agents, a modern code editor, databases, automation, and developer tools.

Automatically checked · GitHub checked 2026-11-16

core-termux

Overview

Core-Termux — Modular Dev Environment

Core-Termux Logo

BUILD. CODE. AUTOMATE.

Version License Platform

Stars Forks Issues Pull Requests

Get Started


CORE-TERMUX is a modular dev environment that turns Termux into a complete development workstation. Through a single core CLI, it provides a modular system that covers the full developer stack: programming languages, databases, AI agents, code editors, shell configuration, and automation — all manageable with simple, consistent commands like core install, core update, and core uninstall.

[!IMPORTANT] This project is designed exclusively for Termux on Android and is not supported on other platforms.


Quick Installation

curl -fsSL https://raw.githubusercontent.com/DevCoreXOfficial/core-termux/main/install.sh | bash

Then run:

core

Main Commands

CommandDescription
core --versionShow current version
core agentLocal AI assistant & task agent
core brainSecond brain — save and search memories
core envManage environment variables
core installInstall specific modules
core showShow tool documentation
core updateUpdate modules or framework
core uninstallRemove installed modules
core reinstallUninstall + reinstall modules
core voiceSpeech-to-agent via microphone
core openOpen documentation in browser
core listList available tools in modules
core pgPostgreSQL database manager
core initConfigure existing projects

Common Modules

These modules are available across most commands (core list, core install, core update, core reinstall, core uninstall, core show, and core open):

ModuleDescription
langLanguage packages (Node.js, Python, Perl, PHP, Rust, C/C++, Go, Bun.js)
dbDatabases (PostgreSQL, MariaDB, SQLite, MongoDB, Redis)
aiAI agents and coding assistants — see AI Agents
editorCode editor components (Neovim, NvChad)
devDevelopment tools (gh, wget, curl, fzf, lsd, bat, etc.)
npmNode.js global npm packages
shellZSH plugins
uiTermux UI components
autoAutomation tools (n8n)

AI Agents

The ai module installs AI-powered coding agents and assistants. Install all agents or pick specific ones with --flag:

core install ai                    # Install all agents
core install ai --opencode --ollama  # Install only OpenCode and Ollama
AgentFlagDescription
Qwen Code--qwen-codeAlibaba's AI coding assistant
Gemini CLI--gemini-cliGoogle's AI assistant with Gemini
Claude Code--claude-codeAnthropic's CLI tool with Claude AI
Mistral Vibe--mistral-vibeCommand-line coding assistant powered by Mistral's models
OpenClaude--openclaudeOpen source Claude Code alternative
OpenClaw--openclawPersonal AI Assistant
Ollama--ollamaRun open-source LLMs locally on Termux
Codex CLI--codexCoding agent from OpenAI that runs locally on your computer
OpenCode--opencodeOpen-source agent that helps you write code in your terminal
Qoder--qoderA terminal-native AI coding partner—and an agent engine you can build on
KiloCode CLI--kilocode-cliThe open source coding agent for building with AI in VS Code, JetBrains, or the CLI
Cactus Needle--cactus-needle26M function-call model for tool-call generation — inference and finetuning from the terminal
KeelCode--keelcodeHosted coding agent for your terminal — inspect a project, edit files, run commands, and verify its work
Kimchi--kimchiTerminal coding agent powered by Kimchi's multi-model orchestration
MiMoCode--mimocodeXiaomi's AI coding agent — fast, local, and open-source
Engram--engramPersistent memory system for coding agents
CodeGraph--codegraphAnalyzes your codebase structure and dependencies
Pi Coding Agent--piMinimal terminal coding harness — adapt Pi to your workflows
Antigravity CLI--antigravity-cliLightweight, terminal-first surface for Antigravity agents
MiniMax CLI--minimax-cliGenerate text, images, video, speech, and music from the terminal
Oh-My-Pi--oh-my-piAutonomous coding agent that learns your codebase and writes production-ready code
Gentle AI--gentle-aiEcosystem, Frameworks, Workflows for AI coding agents
Gentleman Guardian Angel--ggaProvider-agnostic AI code review for every commit
Hermes Agent--hermes-agentThe self-improving AI agent built by Nous Research
Kimi Code--kimi-codeKimi Code CLI — The Starting Point for Next-Gen Agents
Command Code--command-codeThe coding agent that learns your coding taste
Freebuff--freebuffA 100% free coding agent, right from your terminal
Context7--ctx7Live documentation provider for AI coding agents
OpenSpec--openspecSpec-Driven Development framework for AI coding agents
SuperCode CLI--supercodeOpen source SWE agent — free models included, or bring your own
Cline CLI--clineThe open source coding agent in your IDE and terminal
AMP Code CLI--ampcodeAMP — coding agent by Sourcegraph for the terminal
Cursor CLI--cursor-cliDeliver code with agents directly from your terminal
Droid Factory--droid-factoryFactory's AI coding agent — droid CLI for the terminal
Hugging Face CLI--hugging-faceOfficial Hugging Face Hub CLI — download, upload, and manage models, datasets, Spaces, buckets, and Jobs
Cactus Compute--cactusCactus Engine CLI — hybrid edge-cloud AI inference engine for mobile, wearable, and edge devices (cactus-compute)
Walkie--walkieP2P communication CLI for AI agents — encrypted serverless mesh so agents and humans can chat

Detailed Commands

core --version

Display the installed version of Core-Termux.

core --version

Output:

4.27.2

core agent

Local AI assistant and task agent backed by an OpenAI-compatible endpoint (default: gemma-4-e2b-it-cq4 served by Cactus Engine on http://127.0.0.1:8000/v1). ask answers questions with colored markdown; run is a full agent that writes files and runs commands on your machine. If the model server is down, the agent starts cactus in the background (logs → ~/.cache/core-termux/core-agent.log) and stops it when you leave the interactive shell.

core agent ask -p "Explain rsync"                     # One-shot question
core agent run -p "create a backup script"            # One-shot task (files + commands)
core agent ask                                        # Interactive chat shell
core agent run                                        # Interactive agent shell
core agent status                                     # Endpoint/model status
core agent config                                     # Show or edit saved settings

Options (ask & run):

OptionDescription
-p, --prompt <text>Task/question (omit for the interactive shell)
-m, --model <name>Model id (default: gemma-4-e2b-it-cq4)
-u, --endpoint <url>OpenAI-compatible endpoint (default: http://127.0.0.1:8000/v1)
-t, --temperature <n>Sampling temperature (default: 0.3)
--max-tokens <n>Max output tokens (default: 2048)
-w, --workspace <dir>Agent working dir (run mode, default: $PWD)
-n, --max-iterations <n>Agent loop limit (default: 12)
-y, --yesAuto-approve commands (skip y/N prompt)
--planPlan mode (read-only): no file writes, write commands blocked
--buildBuild mode (default): files and commands applied

Files & commands:

  • Type @name in a message to attach a file's contents (live fzf picker while typing)
  • Start a message with ! for shell mode (e.g. !git status) — the output is added to the agent's context
  • Commands from the model run only after your y/N confirmation (-y auto-approves)
  • Press ESC ESC at any prompt to cancel the agent
  • The interactive REPL remembers the conversation and shows [context % · elapsed] after each answer/task
  • Dictate your prompt with /voice (Termux:API)

Interactive slash commands: /help, /model <name>, /endpoint <url>, /temp <n>, /max <n>, /workspace <dir>, /plan, /build, /voice, /clear, /history, /status, /exit

Plan vs Build mode: in Plan mode (--plan or /plan) the agent is read-only — ## File: blocks are ignored, write commands (rm, mv, mkdir, redirections, git commit, package installs...) are blocked, and only read-only commands (ls, cat, sed -n, grep, git status...) run. Use it to explore and get a concrete plan before touching anything, then switch to Build mode (/build) to apply it.

Reading files: a file attached with @ is passed to the agent with its FULL content, so the agent already has it in context and doesn't need to re-read it. A ## File: heading never precedes a command: a command-in-File slip is reinterpreted as a command instead of overwriting your file.

Example session:

$ core agent ask

    you ▸ explain git rebase vs merge

    git rebase rewrites the history of your current branch...

Tip: core agent also runs as a walkie agent — walkie agent <channel> --cli core.


core env

Manage environment variables in your shell rc file (.zshrc or .bashrc). All operations are interactive.

core env                     # Show help
core env set                 # Add or update a variable (value is hidden while typing)
core env unset               # Remove a variable (shows list to choose from)
core env ls                  # List all user-defined variables

Features:

  • Values are hidden with ● when typing (safe for API keys and tokens)
  • Detects existing variables and warns before replacing
  • Removes all definitions of the same variable name
  • Writes to .zshrc if it exists, otherwise .bashrc

Example session:

$ core env set

    ┌─────────────────────────────────────────┐
    │         Set Environment Variable        │
    └─────────────────────────────────────────┘

    ┌─ Variable name
    └─▶ OPENAI_API_KEY

    ┌─ Value for OPENAI_API_KEY
    │  (input will be hidden)
    └─▶ ●●●●●●●●●●●●●●

    ✔ Variable OPENAI_API_KEY set in .zshrc
    • Run: source .zshrc to apply

$ core env ls

    ─────── Environment Variables ───────

    File: .zshrc

    OPENAI_API_KEY              = sk-...
    DATABASE_URL                = postgresql://...

    ──────────────────────────────────────
    2 variable(s) in .zshrc

core brain

Save and search personal learnings and memories — your second brain in markdown files. All operations are local, synced optionally to a private GitHub repo.

core brain                    # Dashboard with stats
core brain init               # Initialize brain directory and GitHub repo
core brain save               # Interactive: save a new memory
core brain search <query>     # Search memories by keywords or tags
core brain ls [category]      # List memories by category
core brain edit               # Edit a memory in your $EDITOR
core brain edit <slug>        # Edit a memory by slug name
core brain delete             # Delete a memory permanently
core brain show <slug>        # View a memory with its relations
core brain reset              # Destroy the entire brain
core brain graph              # Visual map of all connections
core brain skill              # Create an AI skill from memories
core brain relate             # Link two memories interactively
core brain sync               # Push/pull to GitHub private repo

Memory format (AI-consumable markdown):

---
title: React Hook Form + Zod validation
tags: [react, forms, typescript, zod]
created: 2026-06-23
category: frontend
related: [nextjs-server-actions]
---

# React Hook Form + Zod validation

After hours of testing, the combination that worked...

Features:

  • Categorized folders (frontend/, devops/, linux/, etc.) with tags for cross-relations
  • Auto-suggests relations based on shared tags when saving
  • Values hidden with ● when typing for API keys and tokens
  • Syncs to a private GitHub repo via gh for backup across devices
  • Markdown frontmatter consumable by AI agents

Example session:

$ core brain save

    ┌─────────────────────────────────────────┐
    │            Save a New Memory            │
    └─────────────────────────────────────────┘

    ┌─ Title
    └─▶ React Hook Form + Zod patterns

    Existing categories:
    • frontend
    • devops

    ┌─ Category
    └─▶ frontend

    ┌─ Tags (comma separated)
    └─▶ react, forms, zod, typescript

    Write your content below (Ctrl+D to finish, Ctrl+C to cancel):

    After hours testing, the definitive combination...
    [Ctrl+D]

    ✔ Memory saved to frontend/2026-06-23_react-hook-form-zod-patterns.md

core voice

Capture voice from the microphone, review it in nvim, and launch an AI agent.

core voice                    # Show help
core voice <agent>            # Capture → nvim → launch agent
core voice text               # Capture → nvim → print to stdout
core voice !                  # Alias for 'text'

Requirements:

Note: core voice automatically runs termux-api-start before capturing audio to ensure the Termux:API service is running.

Supported agents:

AgentCommand
opencodeopencode run "prompt"
qoderqodercli -p "prompt"
claude-codeclaude -p "prompt"
codexcodex "prompt"
gemini-cligemini -p "prompt"
hermes-agenthermes chat -q "prompt"
kilocode-clikilo run "prompt"
kimi-codekimi -p "prompt"
mimocodemimo run "prompt"
mistral-vibevibe --prompt "prompt"
openclaudeopenclaude --bg "prompt"
pipi -p "prompt"
qwen-codeqwen -p "prompt"
textPrint prompt to stdout

Example session:

$ core voice opencode

    ➜ Listening through the microphone...
    ➜ Review the prompt in nvim, fix mistakes, then save and quit
    ➜ Launching opencode with prompt…

    # opencode opens with the voice-transcribed prompt

core show

Display help documentation for any installed tool. Documentation is loaded from the tool's README.md file in its module directory.

core show                    # Show help
core show <module>           # List all tools in a module
core show <module> --<tool>  # Show specific tool documentation

Examples:

core show ai --opencode      # Show OpenCode documentation
core show db --postgresql    # Show PostgreSQL documentation
core show npm --typescript   # Show TypeScript documentation

Colorized output: If bat is installed, documentation is displayed with syntax highlighting. Otherwise, plain text is shown.


core list

List available tools in a module and their installation status.

core list                     # Show help
core list <module>            # List tools in specific module

All modules from Common Modules are valid targets.


core install

Install individual modules or specific tools within modules.

core install                  # Show help
core install <module>         # Install entire module
core install <module> --tool1 --tool2  # Install specific tools

All modules from Common Modules are valid targets.

Install entire module:

core install ai               # Install all AI tools
core install db               # Install all databases
core install dev              # Install all development tools

Install specific tools:

core install ai --qwen-code --ollama          # Install only Qwen Code and Ollama
core install db --postgresql --sqlite         # Install only PostgreSQL and SQLite
core install dev --gh --fzf --jq              # Install only gh, fzf, and jq
core install npm --typescript --prettier      # Install only TypeScript and Prettier

Tip: Run core list <module> to see all available tools and their flags.


core update

Update modules or the complete framework.

core update                   # Show help
core update <target>          # Update specific target
core update <target> --tool1 --tool2  # Update specific tools
core update core              # Update framework only

In addition to all Common Modules, core update also supports:

TargetDescription
coreCore-Termux framework only

Update entire module:

core update ai               # Update all AI tools
core update db               # Update all databases

Update specific tools:

core update ai --qwen-code --ollama          # Update only Qwen Code and Ollama
core update db --postgresql --sqlite         # Update only PostgreSQL and SQLite
core update dev --gh --fzf --jq             # Update only gh, fzf, and jq

core uninstall

Remove installed modules or specific tools.

core uninstall                # Show help
core uninstall <target>       # Uninstall specific target
core uninstall <target> --tool1 --tool2  # Uninstall specific tools

In addition to all Common Modules, core uninstall supports per-module and per-tool removal. No "uninstall all" — remove only what you need.

Uninstall specific tools:

core uninstall ai --qwen-code --ollama        # Uninstall only Qwen Code and Ollama
core uninstall db --postgresql --sqlite       # Uninstall only PostgreSQL and SQLite
core uninstall dev --gh --fzf                 # Uninstall only gh and fzf

core reinstall

Reinstall modules or specific tools — uninstalls then installs from scratch.

core reinstall                # Show help
core reinstall <target>       # Reinstall specific target
core reinstall <target> --tool1 --tool2  # Reinstall specific tools

In addition to all Common Modules, core reinstall supports per-module and per-tool reinstallation. No "reinstall all".

Reinstall specific tools:

core reinstall ai --opencode --ollama       # Reinstall only OpenCode and Ollama
core reinstall db --postgresql --sqlite     # Reinstall only PostgreSQL and SQLite
core reinstall dev --gh --fzf               # Reinstall only gh and fzf

core open

Open official documentation in browser

core open                     # Show help
core open <target>            # Open official documentation in browser

All Common Modules are valid targets, plus:

TargetDescription
coreCore-Termux documentation
devcorexDevCoreX official website

core pg

PostgreSQL database manager.

core pg                       # Show help
core pg start                 # Start server
core pg stop                  # Stop server
core pg restart               # Restart server
core pg status                # Check status
core pg init                  # Initialize database
core pg create <name>         # Create database
core pg drop <name>           # Drop database
core pg list                  # List databases
core pg shell                 # Open psql console

Features:

  • Automatic data directory detection
  • Support for existing installations
  • Logs in ~/.cache/core-termux/postgresql.log

core init

Configure existing projects with predefined dependencies, folder structure, and tooling. Detects your package manager (npm, pnpm, yarn, or bun) and installs dependencies accordingly.

core init                     # Auto-detect project type and configure
core init <template>          # Configure with specific template

What it does:

  1. Detects package manager — Automatically identifies npm, pnpm, yarn, or bun from lock files or installed binaries
  2. Installs dependencies — Adds optional packages based on your selections (Zustand, React Query, Zod, etc.)
  3. Creates folder structure — Sets up a modular architecture with src/components/, src/services/, src/hooks/, etc.
  4. Generates config files — Creates .prettierrc, .env.example, tsconfig.json, and other project-specific files
  5. Preserves existing scripts — Does not modify package.json scripts, so your dev, build, and start commands stay as your template set them

Available templates:

TemplateDescription
nextNext.js with optional Turbopack, TypeScript, Tailwind CSS
reactReact + Vite with modern structure
nestNestJS with TypeORM and authentication
expressExpress API with TypeScript + TypeORM + migrations

Usage:

cd my-next-app && core init next
cd my-react-app && core init react
cd api && core init express
cd backend && core init nest

Example:

$ cd my-next-app && core init next

──────────────────────────────────────────────────────────────
╭────────────────────────────────╮
│ Configuring Next.js Project    │
╰────────────────────────────────╯
──────────────────────────────────────────────────────────────

    ➜ Package manager detected: pnpm

    ┌─ Configure Turbopack (faster dev/build)? [Y/n]
    └─▶ y

    ┌─ Install Zustand (state management)? [Y/n]
    └─▶ y

    ┌─ Create modular folder structure? [Y/n]
    └─▶ y

─────────────────── Creating folder structure ────────────────
    ✔ Created src/components/ui
    ✔ Created src/components/layout
    ✔ Created src/services
    ✔ Created src/hooks
    ✔ Created src/store
    ✔ Created src/types
    ✔ Created src/config
    ✔ Created src/providers

──────────────────────────────────────────────────────────────
    ✔ Next.js configured!
──────────────────────────────────────────────────────────────

Template Details

Next.js (core init next)

Turbopack & LightningCSS Support:

When running in Termux, core init next offers optional Turbopack support (the native Rust-based bundler for Next.js). If the glibc toolchain is installed (core install npm --turbopack), you can enable Turbopack for faster dev/build times. The installer also adds platform-specific native bindings for LightningCSS and Tailwind CSS.

Installed dependencies:

{
  "dependencies": {
    "axios": "latest",
    "lucide-react": "latest",
    "framer-motion": "latest",
    "sonner": "latest",
    "zod": "latest",
    "react-hook-form": "latest",
    "@hookform/resolvers": "latest",
    "@tanstack/react-query": "latest",
    "zustand": "latest",
    "tailwindcss": "latest"
  },
  "devDependencies": {
    "prettier": "latest",
    "prettier-plugin-tailwindcss": "latest",
    "@next/swc-linux-arm64-gnu": "latest",
    "lightningcss-linux-arm64-gnu": "latest",
    "@tailwindcss/oxide-linux-arm64-gnu": "latest"
  }
}

Configuration:

  • .prettierrc with Tailwind CSS plugin
  • Structure: components/, lib/, hooks/, types/, config/, store/

React + Vite (core init react)

Same dependencies as Next.js (except Next.js-specific configs)

Configuration:

  • .prettierrc with Tailwind CSS plugin
  • Structure: components/, lib/, hooks/, types/, config/, store/, pages/

Express.js (core init express)

Dependencies:

express, pg, typeorm, reflect-metadata
jsonwebtoken, cookie-parser, morgan, cors
bcryptjs, helmet, cloudinary, multer
express-rate-limit, tsconfig-paths, zod

devDependencies:

typescript, ts-node-dev, tsconfig-paths, tsc-alias
@types/node, @types/multer, @types/morgan
@types/jsonwebtoken, @types/helmet
@types/express, @types/cors
@types/cookie-parser, @types/bcryptjs

Scripts added:

{
  "dev": "ts-node-dev --require tsconfig-paths/register --env-file=.env --respawn src/index.ts",
  "build": "tsc && tsc-alias -p tsconfig.json",
  "start": "node dist/index.js",
  "typeorm": "ts-node-dev --require tsconfig-paths/register --env-file=.env ./node_modules/typeorm/cli.js",
  "mg:gen": "npm run typeorm -- migration:generate -d src/database/data-source.ts",
  "mg:create": "npm run typeorm -- migration:create",
  "mg:run": "npm run typeorm -- migration:run -d src/database/data-source.ts",
  "mg:revert": "npm run typeorm -- migration:revert -d src/database/data-source.ts",
  "mg:show": "npm run typeorm -- migration:show -d src/database/data-source.ts"
}

Structure created:

src/
├── app.ts                 # Express configuration
├── index.ts               # Entry point
├── config/
│   └── env.ts            # Environment variables
├── database/
│   ├── data-source.ts    # TypeORM DataSource
│   ├── migrations/
│   └── seeds/
├── entities/
├── controllers/
├── repositories/
├── services/
├── routes/
├── schemas/              # Zod schemas
├── middlewares/
├── types/
└── utils/

Configured files:

  • tsconfig.json with paths (@/*)
  • .env.example
  • src/config/env.ts
  • src/database/data-source.ts (TypeORM)
  • src/app.ts (Express with CORS, helmet, rate-limit)
  • src/index.ts

NestJS (core init nest)

Dependencies:

@nestjs/typeorm, typeorm, pg
@nestjs/jwt, @nestjs/passport
class-validator, class-transformer
bcryptjs, helmet, cloudinary

Language Packages

The lang module installs the following programming languages and runtimes via pkg:

core install lang
Language/RuntimePackageDescription
Node.js LTSnodejs-ltsLong-term support release of Node.js
PythonpythonPython 3 interpreter
PerlperlPerl scripting language
PHPphpPHP interpreter
RustrustRust compiler and Cargo
C/C++clangLLVM C/C++ compiler
GogolangGo programming language
BunbunBun JavaScript runtime

Development Tools

The dev module installs the following development utilities via pkg (or compiled from source where noted):

core install dev
ToolPackageDescription
GitHub CLIghOfficial GitHub command-line tool
WgetwgetFile downloader
CurlcurlHTTP client and transfer tool
LSDlsdModern ls replacement with icons and colors
BatbatModern cat replacement with syntax highlighting
ProotprootChroot alternative for user-space
Ncurses Utilsncurses-utilsTerminal UI manipulation tools
TmatetmateInstant terminal sharing
TmuxtmuxTerminal multiplexer
OpenSSHopensshSSH server and client
CloudflaredcloudflaredCloudflare Tunnel client
Translate Shelltranslate-shellCommand-line translator
html2texthtml2textHTML to plain text converter
jqjqLightweight JSON processor
bcbcArbitrary precision calculator
TreetreeRecursive directory listing
FzffzfCommand-line fuzzy finder
ImageMagickimagemagickImage manipulation suite
ShfmtshfmtShell script formatter
MakemakeBuild automation tool
UdockerudockerRun Docker containers without root
SuperFilespfTerminal file manager with TUI, themes, and hotkeys

Node.js Global Modules

The npm module installs the following global npm packages:

core install npm
PackageCommandDescription
TypeScripttscTypeScript compiler
NestJS CLInestNestJS framework CLI
PrettierprettierCode formatter
Live Serverlive-serverDevelopment server with live reload
LocaltunnelltExpose localhost to the internet
Vercel CLIvercelVercel deployment CLI
MarkservmarkservMarkdown live-preview server
PSQL FormatpsqlformatPostgreSQL query formatter
NPM Check UpdatesncuFind outdated dependencies
NgrokngrokSecure tunnel to localhost
Turbopacknext-turbopackNext.js native bundler (requires glibc toolchain)

Turbopack Installation:

core install npm --turbopack

Note: Turbopack requires the glibc toolchain to run on Termux. When enabled, core init next will configure your project with Turbopack for faster development and build times.


Code Editor

The editor module installs Neovim with a custom configuration based on NvChad.

Installation:

core install editor

Features:

  • Neovim - Fast, extensible code editor
  • NvChad - Modern Neovim configuration
  • GitHub Copilot - AI-powered code completion
  • CodeCompanion - AI chat assistant for code
  • Preconfigured plugins - LSP, autocomplete, syntax highlighting, file explorer, etc.

Included languages:

  • TypeScript/JavaScript
  • Python
  • PHP
  • Perl
  • Rust
  • Lua
  • And more...

For detailed information about the editor configuration, plugins, and usage: → Visit: https://github.com/DevCoreXOfficial/nvchad-termux


UI and Logs

The framework includes a professional logging system with colors, icons, and animations, plus a startup banner with random tips.

Log Functions

log_info "Info message"
log_success "Success message"
log_warn "Warning message"
log_error "Error message"
log_debug "Debug message (requires CORE_DEBUG=1)"

Loading Spinner

Hides shell output while running commands:

LOG_FILE="$CORE_CACHE/install.log"

loading "Installing packages" _install_function

_install_function() {
    pkg install packages -y &>"$LOG_FILE"
}

Separators

separator              # Single line
separator_double       # Double line
separator_section "Title"  # Centered title with line

Boxes

box "Title"
box_large "Large title"
box_with_subtitle "Title" "Subtitle"

Interactive Inputs

# Text input
read_input "Name" VAR_NAME

# Confirmation (y/n)
read_confirm "Continue?" VAR_NAME

# Selection with arrow keys ↑↓
read_select "Environment" VAR_NAME "Dev" "Staging" "Production"

# Hidden input (API keys, tokens, passwords) ●●●
read_secret "Value" VAR_NAME

# Multi-line input (no editor needed)
file=$(read_multiline "# Title")
content=$(cat "$file")
rm -f "$file"

Tables

table_start "Col1" "Col2" "Col3"
table_row "value1" "value2" "value3"
table_end

Banner Tips

Every time you open a new Termux session (or run the banner), Core-Termux shows a random tip to help you discover features you might not know about. Tips cover all modules: installing tools, using core brain, managing databases, voice commands, project initialization, and more.

The tip system:

  • Picks a random tip from a pool of 65+ tips on each session
  • Never shows the same tip twice in a row
  • Covers every module and command in the framework

To refresh the tips pool or customize them, edit core/utils/banner.sh.


Project Structure

core-termux/
├── LICENSE
├── README.md
├── assets
│   ├── fonts
│   │   └── font.ttf
│   └── images
│       └── logo.svg
├── core
│   ├── bin
│   │   └── core
│   ├── cli
│   │   ├── commands
│   │   │   ├── --version.sh
│   │   │   ├── brain.sh
│   │   │   ├── env.sh
│   │   │   ├── init.sh
│   │   │   ├── install.sh
│   │   │   ├── list.sh
│   │   │   ├── pg.sh
│   │   │   ├── reinstall.sh
│   │   │   ├── show.sh
│   │   │   ├── uninstall.sh
│   │   │   ├── update.sh
│   │   │   └── voice.sh
│   │   └── core.sh
│   ├── modules
│   │   ├── ai.sh
│   │   ├── auto.sh
│   │   ├── db.sh
│   │   ├── dev.sh
│   │   ├── editor.sh
│   │   ├── lang.sh
│   │   ├── npm.sh
│   │   ├── shell.sh
│   │   └── ui.sh
│   ├── tools
│   │   ├── ai/
│   │   │   ├── all.sh
│   │   │   ├── qwen-code/
│   │   │   │   ├── install.sh
│   │   │   │   └── README.md
│   │   │   ├── claude-code/
│   │   │   │   ├── install.sh
│   │   │   │   ├── bin/claude
│   │   │   │   └── README.md
│   │   │   ├── opencode/
│   │   │   │   ├── install.sh
│   │   │   │   ├── bin/opencode
│   │   │   │   └── README.md
│   │   │   ├── qoder/
│   │   │   │   ├── install.sh
│   │   │   │   ├── bin/qoder
│   │   │   │   ├── helper/qoder_helper.c
│   │   │   │   └── README.md
│   │   │   ├── freebuff/
│   │   │   │   ├── install.sh
│   │   │   │   ├── bin/freebuff
│   │   │   │   ├── helper/freebuff_helper.c
│   │   │   │   └── README.md
│   │   │   ├── keelcode/
│   │   │   │   ├── install.sh
│   │   │   │   ├── bin/keelcode
│   │   │   │   ├── helper/keelcode_helper.c
│   │   │   │   └── README.md
│   │   │   └── ... (13 tools, each with own directory)
│   │   ├── npm/
│   │   ├── lang/
│   │   ├── db/
│   │   │   ├── all.sh
│   │   │   ├── postgresql/
│   │   │   ├── mariadb/
│   │   │   ├── sqlite/
│   │   │   ├── mongodb/
│   │   │   └── redis/
│   │   ├── editor/
│   │   ├── dev/
│   │   ├── shell/
│   │   ├── ui/
│   │   └── auto/
│   └── utils
│       ├── bootstrap.sh
│       ├── banner.sh
│       ├── colors.sh
│       ├── env.sh
│       ├── log.sh
│       └── version.sh
└── install.sh

Configuration

Environment Variables

export CORE_DEBUG=1    # Enable debug logs

Directories

DirectoryDescription
~/.local/share/core-termux-dataPersistent tool data (codegraph, engram, nvchad)
~/.cache/core-termuxLogs and cache
~/.config/core-termuxUser configuration

Log Files

All processes save logs to:

~/.cache/core-termux/
├── install_lang.log
├── install_db.log
├── install_ai.log
├── install_editor.log
├── install_dev.log
├── install_npm.log
├── install_shell.log
├── install_ui.log
├── install_auto.log
├── postgresql.log
├── last_version_check      # Last update check timestamp
└── new_version             # New version available (if exists)

Automatic Updates

The framework checks for updates automatically:

  • Frequency: Once every 24 hours
  • Impact: None (runs in background)
  • Notification: Shown when running core if new version exists
$ core

── Update Available ─────────────────────────────────

⚠ New version available: 4.27.1 (current: 4.27.0)

➜ Run: core update core to update

To update:

core update core

ZSH Shell

When installing the shell module:

Installed Plugins

PluginDescription
powerlevel10kModern and fast theme
zsh-deferDeferred plugin loading
zsh-autosuggestionsSmart autocompletion
zsh-syntax-highlightingSyntax highlighting
zsh-history-substring-searchHistory search
zsh-completionsAdditional completions
fzf-tabFuzzy navigation in completions
zsh-you-should-useCommand suggestions
zsh-autopairAuto-close parentheses
zsh-better-npm-completionBetter npm completion

Persistent Session

The shell saves the current directory and restores it when opening a new session:

# Session 1
$ cd projects/my-app
$ exit

# Session 2
$ pwd
/data/data/com.termux/files/home/projects/my-app  ← Same directory

Configuration:

  • Saves path to ~/.cache/core-termux/last_dir
  • Automatically restored on startup
  • Falls back to $HOME if directory doesn't exist

Usage Examples

Install specific modules

core install db
core install shell
core install npm

Install specific tools within a module

core list ai                                    # See available AI tools
core install ai --qwen-code --ollama            # Install only Qwen Code and Ollama
core install dev --gh --fzf --jq                # Install only gh, fzf, and jq
core install npm --typescript --prettier        # Install only TypeScript and Prettier

Reinstall

core reinstall ai             # Reinstall all AI agents
core reinstall shell          # Reinstall ZSH + plugins
core reinstall ai --opencode --ollama  # Reinstall specific tools

Configure Next.js project

npx create-next-app@latest my-app
cd my-app
core init next

Manage PostgreSQL

core pg init              # First time
core pg start             # Start
core pg create mydb       # Create database
core pg shell             # Open psql
core pg stop              # Stop

Update

core update core          # Framework only
core update shell         # ZSH plugins only
core update ai --qwen     # Specific AI tool only

Uninstall

core uninstall npm        # Remove Node.js modules
core uninstall ai --ollama   # Remove only Ollama

List available tools

core list ai              # List all AI tools and their status
core list dev             # List all development tools
core list db              # List all databases

Important Notes

  1. Restart Termux: After installing shell or ui, restart Termux to apply changes
  2. Permissions: Ensure you have write permissions in the installation directory
  3. Connection: Some installations require internet connection
  4. Logs: Check ~/.cache/core-termux/ if something fails

License

MIT License