M4ST ROUTER
25-provider LLM failover chain with health checks and automatic cost-based routing. Downtime budget: zero.
I'm Anuj — a solo systems architect from Bareilly, India. I wire 21 MCP servers + a voice-enabled bridge core across 11 free LLM providers into agent systems that run on ₹0 cloud budget. Try clicking the logo.
mode:
Everything below runs on free tiers. No credit card, no fluff — just engineered systems.
Live view of the M4ST rig — 20 real MCP servers from the MAST stack. ★ = nobody else ships this. Uptime simulated client-side; the ₹0 goals are real.
Offline wake word "hey m4stclaw" (Vosk) → Groq Whisper STT → pyttsx3 TTS. Barge-in: the AI stops speaking the moment you start talking. Hinglish-optimized.
On tool failure → AI classifies the error (timeout/args/deps/network) → suggests a fix → retries ×3 with backoff → tries an alternative tool. Healing log — every fix becomes future knowledge.
Paste any API key — provider auto-detected from its prefix and added to the routing chain. 11 providers, 9 task chains, live reload: edit .env, no restart.
One source-of-truth config → Cursor, Windsurf, Claude Desktop, Antigravity, Codex, Manus. Same MCP mesh, same memory, zero re-setup.
Real systems, real uptime, real ₹0 invoices.
25-provider LLM failover chain with health checks and automatic cost-based routing. Downtime budget: zero.
21 MCP servers + self-healing bridge stitched into one orchestration layer — memory, tools, voice, and routing that costs exactly nothing.
Ollama edge stack with 12ms semantic cache. Private by default, fast by design.
Fourteen design variants, one identity. Every frame hand-tuned, every particle deliberate.