Turbo31150/jarvis-mcp-toolkit
JARVIS MCP Toolkit — 88+ handlers for autonomous AI agents | Multi-model consensus | 6 GPUs
Platform-specific configuration:
{
"mcpServers": {
"jarvis-mcp-toolkit": {
"command": "npx",
"args": [
"-y",
"jarvis-mcp-toolkit"
]
}
}
}Add the config above to .claude/settings.json under the mcpServers key.
<div align="center">
[](https://modelcontextprotocol.io) [](https://python.org) [](https://nvidia.com)
88+ MCP handlers for autonomous AI agents on a 6-GPU cluster
</div>
graph LR
Agent[AI Agent] --> MCP{MCP Protocol}
MCP --> DB[Database 15]
MCP --> FS[Filesystem 12]
MCP --> API[API Bridge 18]
MCP --> GPU[GPU Mgmt 8]
MCP --> Voice[Voice 10]
MCP --> Trade[Trading 12]
MCP --> Browser[Browser 8]
MCP --> System[Monitor 5]| Category | Count | Key Operations | |----------|-------|----------------| | Database | 15 | CRUD, search, analytics, backup, migration | | Filesystem | 12 | Read, write, watch, tree, backup, sync | | API Bridge | 18 | REST proxy, WebSocket, MCP relay, auth | | GPU Management | 8 | VRAM, thermal, model load/unload, benchmark | | Voice | 10 | STT (Whisper), TTS, commands, wake word | | Trading | 12 | MEXC, signals, consensus, TP/SL, portfolio | | Browser | 8 | CDP navigate, click, fill, screenshot, scrape | | System | 5 | Health, logs, alerts, metrics, restart |
from jarvis_mcp import MCPServer
server = MCPServer(handlers="all")
server.start(port=8901)
# 88 tools now available via MCP protocolWorks with any MCP-compatible client:
.mcp.jsonsettings.jsoncore.router.dispatcherJARVIS Core · TradeOracle · [WhisperFlow](http
Loading reviews...