Skip to content
Michael Elliott edited this page Apr 4, 2026 · 7 revisions

TITAN — The Intelligent Task Automation Network

TITAN is a premium, autonomous AI agent framework built in TypeScript. It connects to 34 LLM providers, supports 15 messaging channels, and ships with ~149 built-in tools across 91 skills — all with zero native dependencies.

Current Version: 1.1.1


Quick Links

Page Description
Getting Started Installation, first run, onboarding wizard
Architecture System design, module map, data flow
Skills Reference All ~149 built-in tools and how to use them
Channels Discord, Telegram, Slack, WhatsApp, Matrix, Signal, Teams, Google Chat, IRC, Mattermost, Lark/Feishu, Email IMAP, LINE, Zulip
Configuration titan.json schema, environment variables, model aliases, autopilot, training
System Prompt How TITAN instructs AI models — ReAct loop, tool enforcement, sub-agent prompts
Dashboard Mission Control v2 React SPA guide
CLI Commands All CLI commands including create-skill, mcp-server, teams
Security Encrypted vault, prompt injection shield, sandbox modes, audit logging
Roadmap Phase history (Phases 1–22) and future plans

Features at a Glance

  • 34 LLM Providers — 4 native (Anthropic, OpenAI, Google, Ollama) + 30 OpenAI-compatible (Groq, Mistral, OpenRouter, Fireworks, xAI, Together, DeepSeek, Cerebras, Cohere, Perplexity, and more)
  • 15 Messaging Channels — Discord, Telegram, Slack, WhatsApp, Matrix, Signal, MS Teams, Google Chat, WebChat, IRC, Mattermost, Lark/Feishu, Email IMAP, LINE, Zulip
  • ~149 Built-in Tools across 91 Skills — Shell, filesystem, web search, browser automation, CAPTCHA solving, cron, webhooks, email, GitHub, vision, voice, computer use, image generation, PDF parsing, calendar, smart home, data analysis, training, self-improvement, income automation, X/Twitter, and more
  • Mission Control v2 — React 19 SPA (Vite + Tailwind CSS 4) with 17 admin panels, live chat, real-time stats
  • LiveKit WebRTC Voice — Real-time voice with Orpheus TTS
  • MCP Server Mode — Expose tools to other agents via JSON-RPC 2.0 (stdio + HTTP)
  • Sub-Agent Orchestration — Spawn, route, and manage multiple concurrent agents with goal management and self-initiative
  • Autopilot — Cron-based autonomous operation (goals, checklist, self-improve modes)
  • Shared Browser Pool — Playwright-based with CapSolver CAPTCHA solving
  • Self-Improvement System — LLM-as-judge evaluation, autoresearch
  • Local Model Fine-Tuning — LoRA via unsloth, GGUF export, Ollama integration; dual training pipelines (Tool Router + Main Agent)
  • RAG / Vector Search — FTS5 + embeddings for retrieval-augmented generation
  • Token Streaming — SSE + WebSocket real-time streaming
  • Deep Research Agent — Multi-step research with source synthesis
  • Mesh Networking — mDNS + WebSocket auto-discovery, HMAC auth, cross-node task routing
  • Graph Memory — Entity-relationship memory with auto-extraction and context injection
  • Team Mode with RBAC — Owner, admin, operator, viewer roles
  • Encrypted Secrets Vault — AES-256-GCM encryption
  • Prompt Injection Shield — Pattern matching + keyword density analysis
  • ContextEngine Plugin System — Extensible context injection
  • Prometheus Metrics — Full observability
  • Fallback Model Chains — Automatic failover with self-heal
  • Skill Scaffolding CLI — JS/TS/YAML templates via titan create-skill
  • One-Line Installercurl -fsSL .../install.sh | bash + cloud deploy (Railway, Render, Replit)
  • Zero Native Dependencies — Pure ESM TypeScript, runs on any Node.js 20+ system

Stats

Metric Value
Version 1.1.1
LLM Providers 34 (4 native + 30 OpenAI-compatible)
Channels 15
Tools ~149 across 91 skills
Tests 4,655 across 154 files (vitest)
Default Model anthropic/claude-sonnet-4-20250514
Gateway Port 48420
Runtime Node.js >= 20, pure ESM, TypeScript strict

Project Info

  • Author: Tony Elliott (Djtony707)
  • License: MIT
  • npm: titan-agent
  • Node.js: >= 20.0.0

Clone this wiki locally