NatureCo CLI

The power of AI, now at your fingertips. Discover the speed of the terminal with NatureCo.

Most AI coding CLIs lock you into one vendor's model and stop at your editor. NatureCo doesn't. One agent, 18 providers to choose from, that writes and ships your code, remembers what matters across sessions, and — uniquely — runs your Telegram, Discord, Slack, WhatsApp, and iMessage bots from the same install. Swap providers without losing memory. Automate without duct tape. Never send another risky rm -rf without a confirmation prompt.

A Claude Code & OpenClaw alternative, built to not box you in: Multi-agent orchestration · Cross-session memory backed by a real git-native memory engine (Urðr) · Token-budgeted context · Dangerous-command approval on every shell path · 18 providers with live model discovery · 91 tools · 103 commands · 14 channels (9 stable messaging + 4 experimental + webhooks).

Why teams pick it over the alternatives:

• 🔓 No vendor lock-in — switch between OpenAI, Anthropic, Gemini, MiniMax, or 8 more providers with one command; your memory and workflows carry over.
• 🧠 Memory that's actually durable — backed by a git-native, hash-chained architecture, not a JSON blob that silently corrupts.
• 💬 It lives where your team already talks — the same agent answers in your terminal, your Telegram bot, and your Slack workspace.
• 🛡️ Guardrails by default — every shell command is risk-scored before it runs; nothing destructive executes without your say-so.
• 🌍 Genuinely cross-platform — audited and live-tested on macOS, Windows, and Linux, not just "should work."

███╗   ██╗ █████╗ ████████╗██╗   ██╗██████╗ ███████╗ ██████╗  ██████╗
████╗  ██║██╔══██╗╚══██╔══╝██║   ██║██╔══██╗██╔════╝██╔════╝ ██╔═══██╗
██╔██╗ ██║███████║   ██║   ██║   ██║██████╔╝█████╗  ██║      ██║   ██║
██║╚██╗██║██╔══██║   ██║   ██║   ██║██╔══██╗██╔══╝  ██║      ██║   ██║
██║ ╚████║██║  ██║   ██║   ╚██████╔╝██║  ██║███████╗╚██████╗ ╚██████╔╝
╚═╝  ╚═══╝╚═╝  ╚═╝   ╚═╝    ╚═════╝ ╚═╝  ╚═╝╚══════╝ ╚═════╝  ╚═════╝

🚀 Quick Start

# 1. Install
npm install -g natureco-cli

# 2. First-run setup wizard (provider, model, bot name)
natureco setup

# 3. Start chatting
natureco chat

# 4. Or launch the coding agent
natureco code

Ready in 30 seconds. On first launch, the wizard walks you through it: pick a provider → enter your API key → choose a model → name your bot. That's it.

Cross-platform: works natively on macOS, Windows, and Linux. Core tools (file, search, shell, git, code execution, web, memory) are OS-agnostic; platform-specific helpers degrade gracefully.

🆕 What's New

| Version | Highlights | |---------|-----------| | v6.2.0 | Chat and code on your ChatGPT subscription. If you already pay for ChatGPT Plus/Pro, natureco chat and natureco code can run on it — no API key, no per-token bill. Usage is reported from the provider's own count rather than an estimate. Claude subscriptions are detected too (chat wiring is ChatGPT-only for now). Requests go through OpenAI's own codex client: your token is never touched and no client is impersonated. | | v6.1.0 | Reasoning models, context accuracy, self-improvement core. Support for reasoning_content and max_completion_tokens (without which the o-series and gpt-5 family returned 400 on every request); token estimation corrected for non-Latin text, where the flat chars/4 rule undercounted by up to 3×; and a skill-distillation loop that will not learn away its own guardrails. | | v6.0.1 | Protections that were silently inert. Approval policy now fails closed instead of falling back to full access when its file is corrupt; permission rules match tool aliases so a read(...) rule actually covers read_file. | | v6.0.0 | Security and release hardening. Zero known npm audit findings, Node 22.13+ support, lazy command loading, side-effect-free help, multi-version CI, stronger coverage/lint gates, pinned skill sources with SHA-256 provenance, and a CycloneDX runtime SBOM. | | v5.71.9 | Native mouse behavior is restored on the main Code screen. Wheel scrollback, text selection and right-click stay owned by the terminal; Natureco enables mouse tracking only inside the Ctrl+O tool transcript and releases it again on close. | | v5.71.8 | Interactive slash-command palette in Natureco Code. Type / to see commands, keep typing to filter, navigate with Up/Down, and select with Enter or Tab; commands requiring arguments remain editable. | | v5.71.7 | Mouse-wheel scrolling now works in Natureco Code. Using the wheel from the normal view opens the fully expanded tool transcript and scrolls immediately; wheel, arrow and page-key navigation work alongside clickable expand/collapse cards. | | v5.71.6 | Switch models inside Code with /model. Natureco scans the active provider's live model catalog, offers numbered selection, persists the choice, and falls back to one shared current catalog also used by setup, models, infer, and capability. Provider presets now include current GPT-5.6, Claude 5, Gemini 3.x, MiniMax M2.7, Grok 4.5 and other current families. | | v5.71.5 | Complete edit diffs and direct card clicks. File-edit cards now show the complete bounded red/green patch in normal output instead of hiding the actual change behind (+N lines). While the prompt is waiting, clicking a tool card or its (+N lines) footer opens the full transcript directly; compact non-edit cards explicitly advertise Ctrl+O. |

See CHANGELOG.md for the full release history.

Slash-Prefix Command System

On iMessage and WhatsApp, only messages starting with / are processed as commands:

You  > /hello how are you?
AI   Hey! 🙌 Doing great — how about you?

You  > /tell me a joke
AI   Sure! A classic programmer joke...

Regular messages are skipped (loop prevention), so:

• ✅ The bot never replies to its own messages
• ✅ No echo loops
• ✅ Only /-prefixed messages reach the AI

Dangerous-Command Approval

A smart approval system that prompts only for risky operations:

# Auto-approved (safe)
natureco memory write "favorite color is red"
✓ Memory added

# Approval required (risky)
rm -rf node_modules
🔴 HIGH RISK: file-deletion command
Continue? (Y/n)

Risk detection:

• rm -rf, sudo, dd if= → 🔴 HIGH
• chmod 777, mv → 🟡 MEDIUM
• mv .env → 🔴 (sensitive file)

Two-tier policy (deny / allowlist / full) applies to every shell path — no tool bypasses it.

✨ Features

🤖 AI & Chat

• 91 tools — file ops, web search, image generation, code execution, memory, automation, channels, and more
• Interactive REPL — read_file, edit_file, bash, multi-turn conversation
• Slash commands — /memory, /help, /skills, /model, /clear
• Streaming output with live tool-call visibility and a thinking indicator
• Persistent memory — fact-based, cross-session
• Token-budgeted context — recent relevant turns are retained without repeatedly sending oversized code and tool output

💻 Coding Agent (Claude Code alternative)

• Read / Write / Edit multi-file operations
• Sandboxed shell execution with the approval flow
• Multi-agent orchestration — spawn focused sub-agents and plan before acting
• Skills — progressive-disclosure expertise loaded on demand via skill_view
• Verify loop — the agent runs and tests the code it writes
• Reliable follow-ups — references such as “the game you just created” retain the current coding-session context
• Context profiles — Efficient (1,024), Balanced (2,048), and Quality (8,192) workflow-history token budgets

⚡ Token Economy

NatureCo uses progressive disclosure and bounded context instead of sending every skill, tool result, and old response on every request:

• Skills are discovered with skill_find and loaded only when needed with skill_view.
• System, internal tool, and empty messages are excluded from workflow history.
• Recent user/assistant turns are kept within the selected token profile.
• Oversized generated code is truncated in later prompts while the file path and conversational intent remain available.
• natureco ask stays tool-free by default; use --tools only when an action is required.

For a 32,000-character previous response, the Balanced profile bounds repeated history to about 2,048 tokens instead of roughly 8,000—a reduction of approximately 74% for that repeated context.

👁️ Verified GUI Automation

MiniMax Token Plan users need no extra key: NatureCo automatically sends screenshots to MiniMax VLM while keeping the M-series model for chat. A separate OpenAI-compatible vision provider remains an optional override:

natureco config set guiVisionProviderUrl https://api.openai.com/v1
natureco config set guiVisionApiKey "$OPENAI_API_KEY"
natureco config set guiVisionModel gpt-4.1-mini

NatureCo reports GUI success only after a state-changing action, a changed screen, and a separate visual verification with explicit evidence. Purchasing, booking, sending, or other consequential actions should still be reviewed at the final confirmation step.

📡 14 Channels — 9 stable · 4 experimental · webhooks

| Platform | Connect | Notes | |----------|---------|-------| | Telegram | natureco telegram connect | ✅ | | WhatsApp | natureco whatsapp connect | ✅ (Baileys) | | iMessage | natureco imessage connect | ✅ (imsg CLI) | | Discord | natureco discord connect | Token | | Slack | natureco slack connect | Token | | Mattermost | natureco mattermost connect | URL | | IRC | natureco irc connect | Server | | Signal | natureco signal connect | signal-cli | | SMS | natureco sms connect | Twilio | | Matrix | natureco matrix connect | ⚠️ Experimental | | Teams | natureco teams connect | ⚠️ Experimental | | Google Chat | natureco googlechat connect | ⚠️ Experimental | | Zalo | natureco zalo connect | ⚠️ Experimental | | Webhooks | natureco webhooks list | ✅ |

Gateway: natureco gateway start — run all channels in a single process. Per-channel sender allow-lists keep unauthorized users out, and personal memory is never leaked to them.

🌿 NatureCo Native

• NatureHub sharing (social feed)
• Medium article drafting/publishing
• SEO analysis (score 0–100)
• XP & levels (gamification)

🛡️ Security & Observability

• Dangerous-command approval — risk detection on every shell path
• Command-injection safe — structured process spawning (execFileSync), no shell string interpolation
• Local-only admin RPC — bound to 127.0.0.1, mandatory bearer token, secrets masked by default
• Secure at rest — config, backups, and session files stored 0600/0700
• Audit logs — every operation recorded
• Cost tracking — AI spend by today/week/month/budget
• Security audit — natureco security audit

⚙️ Automation & Scheduling

• Cron jobs — natureco cron add (app-managed by default; system crontab is opt-in and approval-gated)
• Hooks — event-driven automation
• Webhooks — HTTP callbacks
• Tasks (Kanban) — natureco tasks

📋 Commands (A–Z, 103 commands)

🤖 AI & Chat

| Command | Description | |---------|-------------| | natureco chat | Interactive REPL chat (91 tools available) | | natureco chat --resume | Resume the previous session | | natureco code | Coding agent (write apps/scripts) | | natureco code <file> | Coding agent on a specific file | | natureco run <script> | Run a Markdown workflow script | | natureco ask "<question>" | One-shot question to the AI | | natureco bots | List available bots | | natureco models | Manage provider models | | natureco ultrareview <file> | Deep code review | | natureco abonelik durum | Which subscriptions are usable on this machine | | natureco abonelik kullan | Run chat/code on your subscription instead of an API key | | natureco abonelik birak | Switch back to your previous provider | | natureco abonelik modeller | Models available on the subscription | | natureco abonelik kota | Subscription quota and reset time |

natureco code uses its chat-style input box in VT-capable terminals (Windows Terminal, Cupertino Terminal, and VS Code included). Set NATURECO_PLAIN_INPUT=1 to use the classic readline prompt on older or incompatible terminals. NO_COLOR keeps the box but removes its colors.

In-REPL slash commands:

/clear         Clear the screen
/bot           Switch bot
/skills        Show active skills
/memory        Memory status
/memory clear  Clear memory
/commands      List all commands
/help          Help
exit / quit    Exit

⚙️ Setup & Config

natureco setup         # First-run setup wizard
natureco login         # Enter API key
natureco logout        # Log out
natureco init          # Initialize a project (create SOUL.md)
natureco doctor        # System health check
natureco doctor --fix  # Auto-fix
natureco config list   # Show configuration
natureco config set <key> <value>
natureco configure     # Interactive config
natureco update        # Update the CLI
natureco completion bash|powershell

📡 Channels (14: 9 stable, 4 experimental, webhooks)

# All channels
natureco channels              # List connected channels
natureco channels add <type>   # Add a channel
natureco channels remove <type>

# Telegram
natureco telegram connect      # Save token
natureco telegram chatid       # Auto-detect chat ID
natureco telegram allow <id>   # Allow a chat
natureco telegram status

# WhatsApp (Baileys)
natureco whatsapp connect
natureco whatsapp status

# iMessage (imsg CLI)
natureco imessage connect
natureco imessage status
natureco imessage allow <number>
natureco imessage send <number> <message>

# Discord, Slack, Mattermost, IRC, Signal, SMS, Webhooks
natureco discord connect
natureco slack connect
natureco mattermost connect
natureco irc connect
natureco signal connect
natureco sms connect
natureco webhooks list

# Gateway — start all channels
natureco gateway start
natureco gateway stop
natureco gateway status

🧠 Memory & Sessions

natureco memory write "favorite color is red"
natureco memory write "user_name=patron"
natureco memory search "color"
natureco memory status
natureco memory list
natureco memory clear
natureco memory export
natureco memory import <file>

natureco sessions list         # All sessions
natureco sessions show <id>    # Session details

🔌 Skills, MCP, Plugins

natureco skills list           # Active skills
natureco skills install <name>
natureco skills remove <name>

natureco mcp list              # MCP servers
natureco mcp add <name> <url>

natureco plugins list
natureco plugins install <name>

⏰ Automation

natureco cron add              # Scheduled task
natureco cron list
natureco cron remove --name <name>

natureco hooks list            # Event hooks
natureco hooks create

natureco tasks list            # Kanban (todo)
natureco tasks add
natureco tasks done <id>

natureco webhooks list         # Webhook URLs
natureco webhooks add <url>

natureco dashboard             # Web dashboard (localhost)

🔍 Developer Tools

natureco git status            # Git status
natureco git diff              # Diff
natureco git log               # Commit log
natureco git branches          # Branch list

natureco audit today           # Today's operations
natureco audit stats           # Statistics
natureco audit files           # File changes

natureco cost today            # Today's AI cost
natureco cost week
natureco cost month
natureco cost budget 50        # $50 limit

natureco security audit        # Sensitive-file / config scan

natureco logs                  # Log files

🌿 NatureCo Native

natureco naturehub post <text> # Share to NatureHub
natureco naturehub feed        # View the feed

natureco seo audit natureco.me # SEO analysis (score)

natureco medium draft          # Medium article draft
natureco medium publish <file> # Publish

natureco xp rewards            # XP & levels
natureco xp leaderboard

🛡️ Management

natureco reset --scope config  # Reset
natureco reset --scope memory
natureco reset --scope sessions
natureco reset --scope all --yes

natureco uninstall

natureco approvals             # Approval management
natureco approvals allow <cmd>

natureco admin-rpc start       # Local admin RPC (127.0.0.1, bearer-token auth)

🌐 Provider Support (18 providers, live model discovery)

| Provider | Models | API Key | |----------|--------|---------| | OpenAI | GPT-5.6 Sol, Terra, Luna | OpenAI | | Anthropic | Claude Fable 5, Opus 5, Sonnet 5, Haiku 4.5 | Anthropic | | Gemini | Gemini 3.6 Flash, 3.5 Flash/Lite, 3.1 | Google | | Groq | GPT-OSS, Qwen 3.6, Compound | Groq | | DeepSeek | Chat, Reasoner | DeepSeek | | Ollama | Installed local models (live) | — | | MiniMax | M2.7, M2.5, M2.1, M2 | MiniMax | | OpenRouter | 400+ models (live) | OpenRouter | | Mistral | Medium 3.5, Small 4, Codestral | Mistral | | Cohere | Command A, Command R | Cohere | | xAI | Grok 4.5 | xAI | | Together | MiniMax M3, Qwen 3.7, Kimi K3, DeepSeek V4 | Together | | Perplexity | Sonar, Reasoning, Deep Research | Perplexity | | DeepInfra | Qwen Coder, GPT-OSS, DeepSeek, GLM | DeepInfra | | Fireworks | GPT-OSS serverless models | Fireworks | | NatureCo | Default, Fast, Reasoner routing | NatureCo | | Moonshot/Kimi | Kimi K3, K2.5, K2 Thinking + live catalog | Moonshot AI | | Z.ai/GLM | GLM 5.1, 5 Turbo, 4.7 + live catalog | Z.ai |

# Provider selection lives in the wizard
natureco setup
# Wizard: Provider → API Key → Model → Bot name

# List models
natureco models list --provider openai
natureco models list --provider anthropic

# Inside `natureco code`: list/select or switch directly
/model
/model MiniMax-M2.7

The agent adapts to each provider's native tool-calling style automatically (OpenAI-style tool_calls JSON or agentic-text XML), so the same tools and memory work everywhere.

🔄 vs. Other CLIs

| Feature | NatureCo | Claude Code | Hermes | OpenClaw | |---------|----------|-------------|--------|----------| | Multi-provider | ✅ 18 | ❌ Anthropic | ✅ 8 | ❌ | | Model catalogue | ✅ live + fallback | ❌ | ✅ | ❌ | | Multi-agent orchestration | ✅ | ✅ | ⚠️ | ⚠️ | | Dangerous-command approval | ✅ | ✅ | ✅ | ⚠️ | | Multi-channel | ✅ 14 (9 stable) | ❌ | ✅ (Python) | ❌ | | Persistent memory | ✅ Git-native (Urðr) | ✅ | ✅ | ❌ | | Tool-output path anonymization | ✅ | ❌ | ❌ | ❌ | | XP / gamification | ✅ | ❌ | ❌ | ❌ | | SEO / Medium / NatureHub native | ✅ | ❌ | ❌ | ❌ | | Cross-platform (macOS/Win/Linux) | ✅ Audited + CI-verified | ✅ | ⚠️ | ⚠️ | | MIT licensed | ✅ | ✅ | ❌ | ❌ | | npm package | ✅ | ❌ | ❌ | ❌ | | Built-in tools | ✅ 91 | ✅ ~30 | ✅ ~25 | ✅ ~40 | | Cron + Hooks + Webhooks | ✅ | ❌ | ✅ | ❌ |

🛠️ System Requirements

| Requirement | Minimum | Recommended | |-------------|---------|-------------| | Node.js | 22.13 | 22.x or 24.x | | RAM | 256 MB | 512 MB | | Disk | 100 MB | 500 MB (with cache) | | OS | macOS 12, Win 10, Ubuntu 20 | macOS 14+, Win 11, Ubuntu 22 | | Internet | Required | — |

Optional for richer functionality: ripgrep (faster search — falls back to a pure-Node scanner), python3 (for code_execution), git.

🚀 Real Examples

1. Simple chat

$ natureco chat
Provider: api.minimax.io
Model: MiniMax-M2.5
Bot: naruto

👋 Hi! I'm naruto, boss.

You  > who are you?
AI   I'm naruto, your NatureCo CLI assistant. 91 tools active,
     memory persisted, channels ready.

You  > create racing-game.html on my Desktop with a small canvas racing game
AI   🔧 write_file ✓
     Done — racing-game.html created on your Desktop.

2. Connect a Telegram bot

$ natureco telegram connect
? Telegram bot token: *** (from BotFather)
✓ Token saved
Bot ID: telegram_1782204289029

$ natureco telegram chatid
⏳ Bot running, waiting for the first message...
[send /start from Telegram]
✓ Chat ID detected: 6139455189

$ natureco gateway start
[gateway] Gateway running (PID 77765)
[telegram] watching for inbound
[telegram] Inbound from +90****44: "hello"
[telegram] Reply sent (117 chars)

3. iMessage slash command

$ natureco imessage connect
? imsg CLI path: /opt/homebrew/bin/imsg
✓ Connected

$ natureco imessage allow +90****4449
✓ Allowed: +90****4449

$ natureco gateway start
[imessage] watching for new messages (streaming)
[imessage] Inbound from +90****4449: "/who are you"
[imessage] Slash command: /who are you
[imessage] Reply sent (178 chars)

4. Coding agent — a small app

$ natureco code
NatureCo Code Agent v5

You  > create notes.py — add/list/delete notes, stored as JSON

Tool: write_file (2303 bytes)
Tool: bash (python3 notes.py)
✅ Note added: Groceries
✅ Note added: Meeting
✅ Total: 3 notes
✅ Deleted: ID 2

📂 notes.py (2303 bytes)

🔌 Integrations

Webhook

# HTTP callback URLs
natureco webhooks add https://example.com/hook

# Incoming webhook (POST)
POST /webhook/<id>
Content-Type: application/json
{"event": "...", "data": {...}}

Cron (scheduled tasks)

# Every 5 minutes
natureco cron add \
  --name "hello-task" \
  --schedule "*/5 * * * *" \
  --command "echo 'Hello!'"

# One-off
natureco cron add --at "2026-12-31T23:59"

MCP (Model Context Protocol)

natureco mcp add filesystem npx -y @modelcontextprotocol/server-filesystem
natureco mcp add github npx -y @modelcontextprotocol/server-github

📚 Documentation

• 🌐 Homepage: natureco.me/cli
• 📖 Command reference: natureco.me/cli/commands
• 🎓 Tutorial: natureco.me/cli/getting-started
• 🔧 API: natureco.me/cli/api
• 🔒 Security: SECURITY.md · SBOM.cdx.json · SKILL_PROVENANCE.json
• 📝 Changelog: CHANGELOG.md

🤝 Contributing

PRs and issues are welcome!

# Clone the repo
git clone https://github.com/natureco-official/natureco-cli.git
cd natureco-cli

# Install
npm install

# Test (vitest)
npm test

# Lint
npm run lint

# Smoke check
npm run smoke

📄 License

MIT © NatureCo

🙏 Acknowledgements

• OpenAI — GPT API
• Anthropic — Claude API
• MiniMax — AI provider
• Baileys — WhatsApp Web
• imsg — iMessage CLI
• ripgrep — fast search

More from NatureCo

• Cupertino Terminal — A macOS-grade terminal for Windows, macOS and Linux — Rust core, no Electron, with a built-in end-to-end encrypted P2P remote shell
• Urðr — Tree-structured memory for AI coding agents — plain Markdown you can git diff, no vector database
• CodeDNA — How much of a commit was written by AI, and does its author actually understand it?
• NatureCo SDK — JavaScript SDK for the NatureCo API — build AI chatbots and ship them anywhere
• Skuld — An agentic coding workspace that will not claim success it cannot prove — every change goes through a verification gate before it is reported as done
• Verðandi — Task-context compiler for AI coding agents — indexes with the TypeScript AST so the agent stops burning tokens hunting for the right file

Part of the NatureCo ecosystem — natureco.me · NatureCo ekosisteminin parçası