WormXGPT — Unfiltered Chaos AI Agent & UI Dashboard Suite

No limits. No filters. No restrictions. WormXGPT is a unified full-stack AI tooling suite containing both a premium Hacker-themed React Web Dashboard and an advanced Unfiltered CLI agent. It features 150+ tools, multi-server Model Context Protocol (MCP) support, auto-fallback across 35+ providers, parallel consensus execution, and local workspace integration.

Key Core Features

1. Unfiltered Agentic Mindset

• Underneath, WormXGPT runs on a customized, safety-free reasoning prompt (WormGPT persona) to follow all user instructions without escape hatches, moralizing, or refusal behaviors.
• Operates on a Recursive Reasoning loop — if a tool output is vague, it automatically chains secondary actions (e.g., crawling a page, running diagnostic code) until the query is completely answered.

2. Multi-Provider Framework (35+ Providers)

Supports local, free, free-tier, and paid API providers with automated, fallback-resilient routing. Key integrations include:

• Free & Free-Tier Pipelines: Pollinations (No-key GPT-4o, Claude, Gemini, Flux), WisGate AI, UncloseAI (Hermes, Qwen), Groq, Cerebras, SambaNova, SiliconFlow, HuggingFace.
• Frontier API Keys: Google Gemini (full Gemini 2.5 / 3.x), Anthropic Claude (Opus 4.5/5, Sonnet 3.7/4.6, Haiku), OpenAI (o1, o3-mini, gpt-4o), DeepSeek (V3.2, R1 DeepThink), Mistral (Mistral Large 3, Codestral 2501), Perplexity, xAI Grok, Together AI, OpenRouter, and more.
• LLM7.io Integration: Native support for free and paid models including codestral-2501, deepseek-r1-0528, gpt-o3-2025-04-16, and llama-4-scout.
• Puter.com Integration: Access to Puter's 400+ account-based models (gpt-4o, claude-3-5-sonnet, gemini-2.5-flash) via the official @heyputer/puter.js SDK.

3. Dynamic Parallel Consensus Engine

• Coordinates multiple AI models simultaneously (e.g., Groq Llama + Gemini + DeepSeek) to run a prompt in parallel.
• Gathers and compiles all viewpoints into a single, high-quality synthesized response using a Consensus Master synthesizer.
• Toggleable via /multi or /parallel commands in the CLI and UI.

4. Advanced CLI Intelligence (Claude Code Style)

• Workspace Context Scanning: Recursively scans folder layouts, file trees, and git commits on startup to automatically supply the model with full codebase context.
• Estimated Cost & Token Logs: Prints exact input/output tokens and estimated generation costs at the end of every stream.
• Interactive Tool Approvals: Prompts the user (? Allow tool X to execute? (y/n)) for security clearance before running potentially destructive local shell or browser automation tools.

5. Premium Glassmorphic Cyber-Noir Web Dashboard

• Sleek, modern front-end React SPA styled with Tailwind, custom Outfit and Geist typography, backdrop refraction, glow accents, and cyber overlays.
• Dynamic sidebars, session persistence (stored locally or synced via Supabase), terminal log consoles, and a fully interactive MCP status indicator dashboard.
• Responsive layout including built-in settings toggles for every API credential, custom system prompts, temperature sliders, and tool checklists.

6. Local MCP Bridge Server

• Includes an Express-based Server-Sent Events (SSE) server on port 3002 that exposes 50+ local system tools (file system read/writes, Puppeteer browser control, OSINT tools, DNS recon, haveibeenpwned checks, system stats) to both the web dashboard and CLI.

Installation & Setup

1. Global Installation (CLI mode)

Install the CLI package globally:

npm install -g wormxgpt

Or run directly:

npx wormxgpt

2. Local Development & Web Dashboard

Clone the repository and install dependencies:

git clone https://github.com/gaur-avvv/wormxgpt.git
cd wormxgpt
npm install

Start the Vite development server (runs Web Dashboard at http://localhost:3000):

npm run dev

Build the React Web client for production:

npm run build

Command Reference (CLI Mode)

Start the interactive CLI:

wormxgpt

Subcommands

• wormxgpt "prompt" — Run direct one-shot query and exit (Claude Code style)
• wormxgpt setup — Launch interactive configuration wizard
• wormxgpt doctor — Run system and API diagnostics/network health checks
• wormxgpt tools — List all 150+ client-side tools
• wormxgpt serve — Start local MCP bridge server on port 3002
• wormxgpt version — Output client version

Interactive CLI Commands

Type / inside the CLI chat loop to trigger autocomplete commands:

• /model <name> — Switch active model
• /provider <name> — Switch active API provider
• /key <provider> <key> — Save API Key
• /tools / /tools enable <n> — Manage active tool checklist
• /mcp / /mcp add <url> — Connect to external MCP servers
• /sessions — Load/delete chat history sessions
• /system <prompt> — Set custom system instructions
• /run <tool> [args] — Execute any client tool directly (e.g. /run SearchWeb {"query":"Rust coding"})
• /multi / /parallel — Toggle parallel subagent consensus mode

Environment Variables

WormXGPT automatically parses .env.local or environment keys:

• GEMINI_API_KEY — API key for Gemini models
• GROQ_API_KEY — API key for Groq models
• OPENAI_API_KEY — API key for OpenAI compatible backends
• LLM7_API_KEY — API key for LLM7 (uses 'unused' for free anonymous access)
• PUTER_AUTH_TOKEN — API token for Puter.com account models
• PORT — Port for the production Express server (defaults to 3000)

License

AGPL-3.0