AGENT-X
Your AI Wingman

Multi-provider AI agent in your terminal. 80+ tools. Session persistence. Telegram remote control.
One command to launch. Zero configuration required.

Install · Features · Commands · Providers

---

Overview

Agent-X is an autonomous AI agent that lives in your terminal. It connects to multiple AI providers, wields 80+ built-in tools, remembers context across sessions, and can be commanded remotely via Telegram — all wrapped in a deep-space-themed interface that makes every interaction feel like commanding a starship.

No cloud accounts. No subscriptions. Bring your own API keys and launch.

---

Installation

macOS / Linux:

curl -fsSL https://raw.githubusercontent.com/SlashpanOrg/agent-x/main/install.sh | bash

Windows (PowerShell):

powershell -c "irm https://raw.githubusercontent.com/SlashpanOrg/agent-x/main/install.ps1 | iex"

Requirements: Node.js >= 20 (the only prerequisite). The installer handles everything else.

After installation:

agentx

A guided setup wizard walks you through provider selection, API key configuration, and preferences on first run.

---

Uninstall

curl -fsSL https://raw.githubusercontent.com/SlashpanOrg/agent-x/main/uninstall.sh | bash

---

Features

Multi-Provider AI

Switch between providers mid-conversation. No restart needed.

Provider	Models
OpenAI	GPT-4o, o1, o3
Anthropic	Claude 3.5, Claude 4
Google	Gemini
Ollama	Any local model
LM Studio	Any local model

80+ Built-in Tools

Agent-X does not just talk — it acts. Tools are organized by domain:

• Filesystem — read, write, move, delete, search files and directories
• Shell — execute commands, manage background processes
• Git — status, diff, log, commit, branch, stash, blame
• Code Intelligence — search symbols, find definitions, replace, refactor
• Packages — install, remove, list, outdated, run scripts
• Testing — run suites, watch mode, coverage, generate tests
• Web / HTTP — GET, POST, scrape, search the web
• Browser Automation — open pages, click elements, screenshots, evaluate JS
• Containers — Docker lifecycle, compose, logs, exec
• Database — query, inspect schema, export data
• GitHub — issues, PRs, repos, workflows, releases
• System — disk, ports, env, processes, security audit
• MCP — connect to any Model Context Protocol server for extended capabilities

Permission System

Every tool action passes through a clearance gate:

• Scope-based path validation
• Risk-level assessment per tool
• Interactive prompts — allow once, allow always, or deny
• Full audit trail of approved actions

Session Persistence

• Auto-save on every turn
• Crash recovery — pick up exactly where you left off
• Token tracking and context management
• Session compaction when context grows large

Telegram Integration

Run Agent-X as a background daemon and interact with it remotely through Telegram. Shared session context, permission prompts forwarded to your phone, and full command support — your agent is always reachable.

agentx start      # launch the background daemon
agentx status     # check daemon health
agentx stop       # terminate the daemon

Profiles

Create multiple agent personalities — each with its own system prompt, tone, and behavioral boundaries. Switch between a DevOps specialist, a writing assistant, and a code reviewer without reconfiguring anything.

Sub-Agents

Delegate complex subtasks to lightweight satellite agents with isolated tool access. The main agent orchestrates; sub-agents execute and report back.

---

Commands

All configuration and control happens inside the Agent-X terminal:

Command	Description
/help	Show all available commands
/model <name>	Switch AI model
/provider <name>	Switch provider
/config	Manage settings and API keys
/profile	Switch or create profiles
/session list	List saved sessions
/session restore <id>	Restore a previous session
/clear	Clear message history
/compact	Summarize and compress context
/permissions	Review and manage tool permissions
/telegram setup	Configure Telegram bot integration
/telegram status	Check Telegram connection
/exit	Enter cryo-sleep

---

Providers

Agent-X works with any OpenAI-compatible API. Configure multiple providers and switch between them freely:

/provider openai
/model gpt-4o

/provider anthropic
/model claude-sonnet-4-20250514

/provider ollama
/model llama3

Local models via Ollama and LM Studio require no API key — just a running server.

---

Supported Platforms

Platform	Architecture
macOS	Apple Silicon (arm64)
macOS	Intel (x64 via Rosetta)
Linux	x64
Linux	arm64
Windows	x64

---

Version Pinning

Install a specific version:

AGENTX_VERSION=v0.1.0 curl -fsSL https://raw.githubusercontent.com/SlashpanOrg/agent-x/main/install.sh | bash

---

Philosophy

Agent-X is built on three principles:

1. Autonomy with accountability — The agent acts independently but never bypasses your clearance. Every destructive or sensitive action requires explicit approval.

2. Local-first — Your data stays on your machine. Sessions, memories, and configuration never leave your system. Nothing phones home.

3. Provider-agnostic — No lock-in. Swap between cloud and local models at will. The same tools and workflows work regardless of the AI backend.

---

Ground control to Major — systems nominal.