Getting Started with SuprMass

01What is SuprMass

SuprMass is a distributed AI operating system built on a partner mesh of world-class compute providers. Instead of asking a single model, you ask a council of 10 specialized agents — each with hyperdirect memory access, operating in parallel across the stack.

Traditional systems lock you to a training cutoff, markup costs 3–10x, and route everything through a single monolithic instance. SuprMass breaks from that model entirely:

• Distributed partner mesh — inference cascades across specialized nodes
• Hyperdirect memory — zero-latency access to context and state
• Future-state inference — not locked to training dates
• At-cost pricing — no markup on compute
• Deterministic skills — zero-token local tools for common tasks

The result is a system that moves at a different axis entirely — not incremental speed, but a step change in how AI integrates with your workflow.

02Quick Start

IDE Access

The fastest way to experience SuprMass is through the Sanctum IDE:

1. Open ide.suprmass.com
2. Select your mode: Debug & Fix, Guide Me, or Solo
3. Type a message or pick a quick-action chip
4. The Council responds across the partner mesh

API Access

All endpoints are available through the unified API gateway:

curl -X POST https://api.suprmass.com/council/chat \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [{"role":"user","content":"Refactor this function"}],
    "mode": "fix"
  }'

Or query the agent directly with model routing:

curl -X POST https://api.suprmass.com/agent/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Write a React component",
    "mode": "auto",
    "model": "QWEN"
  }'

03The 10 Workers

The Council is composed of 10 specialized agents, each backed by a world-class model optimized for its domain. When you send a message, the mesh routes it to the right council member — or runs multiple in parallel for complex tasks.

04The Skill System

Skills are deterministic, zero-token tools that run locally in the IDE. They do not call external models — they execute pure logic on your code instantly, at zero cost.

This means common tasks like code review, test generation, and security scanning happen in milliseconds with absolutely no token burn.

Using Skills

In the IDE, click ⚡ Skills in the top bar to open the skills drawer. Click any skill to run it on the currently open file. Results appear in chat with millisecond timing.

You can also trigger skills via terminal commands:

$ skill review   # run code review on current file
$ skill test     # generate tests
$ skill sec      # security scan

05The 3 Modes

SuprMass adapts its behavior based on how you want to work. Switch modes instantly from the top bar in the IDE.

Mode API Parameter

When calling the API, pass the mode to influence council behavior:

{
  "messages": [...],
  "mode": "fix"    // or "guide", "solo"
}

06Command Reference

Keyboard Shortcuts

Terminal Commands

Quick Chips

Click any chip below the chat input for one-click actions:

• 🔍 Review — Run code review
• 🧪 Tests — Generate unit tests
• 📖 Explain — Explain the current code
• ⚡ Refactor — Refactor for performance
• 📝 Docs — Add JSDoc comments
• 🛡️ Security — Check for security issues

07API Endpoints

All requests route through the unified gateway at https://api.suprmass.com. The gateway proxies to specialized backend workers with CORS enabled.

Response Format

All endpoints return JSON with CORS headers. Error responses include error and message fields.

{
  "response": "The Council has analyzed your request...",
  "session_id": "550e8400-e29b-41d4-a716-446655440000",
  "model": "ARCHITECT",
  "mode": "fix"
}

{
  "error": "Not found",
  "path": "/unknown",
  "available_routes": ["/council", "/agent", "/models", ...]
}