Nefesh

Real-time human state awareness for AI agents. Fuses biometric signals into a unified stress score (0-100) via Streamable HTTP.

GitHub

Nefesh MCP Server

A Model Context Protocol server that gives AI agents real-time awareness of human physiological state — stress level, confidence, and behavioral adaptation prompts.

What it does

Your AI agent sends sensor data (heart rate, voice, video, text) via the Nefesh API. The MCP server returns a unified stress score (0–100), a state label (Calm → Acute Stress), and an adaptation prompt that tells the agent how to adjust its behavior.

Signals supported: cardiovascular (HR, HRV, RR intervals), vocal (pitch, jitter, shimmer), visual (facial action units), textual (sentiment, keywords)

Setup

1. Get an API key

Get your key at nefesh.ai/pricing ($25/month, 50,000 calls).

2. Add to your AI agent

Find your agent's MCP config file:

AgentConfig file
Cursor~/.cursor/mcp.json
Windsurf~/.codeium/windsurf/mcp_config.json
Claude Desktop~/Library/Application Support/Claude/claude_desktop_config.json
Claude Code.mcp.json (project root)
VS Code (Copilot).vscode/mcp.json or ~/Library/Application Support/Code/User/mcp.json
Clinecline_mcp_settings.json (via UI: "Configure MCP Servers")
Continue.dev.continue/config.yaml
Roo Code.roo/mcp.json
Amazon Q~/.aws/amazonq/mcp.json
JetBrains IDEsSettings → Tools → MCP Server
Zed~/.config/zed/settings.json (uses context_servers)
OpenAI Codex CLI~/.codex/config.toml
Goose CLI~/.config/goose/config.yaml
ChatGPT DesktopSettings → Apps → Add MCP Server (UI)
Gemini CLISettings (UI)
AugmentSettings Panel (UI)
ReplitIntegrations Page (web UI)
LibreChatlibrechat.yaml (self-hosted)

Add the following configuration (works with most agents):

{
  "mcpServers": {
    "nefesh": {
      "url": "https://mcp.nefesh.ai/mcp",
      "headers": {
        "X-Nefesh-Key": "<YOUR_API_KEY>"
      }
    }
  }
}
VS Code (Copilot) — uses servers instead of mcpServers 
{
  "servers": {
    "nefesh": {
      "type": "http",
      "url": "https://mcp.nefesh.ai/mcp",
      "headers": {
        "X-Nefesh-Key": "<YOUR_API_KEY>"
      }
    }
  }
}
Zed — uses context_servers in settings.json 
{
  "context_servers": {
    "nefesh": {
      "settings": {
        "url": "https://mcp.nefesh.ai/mcp",
        "headers": {
          "X-Nefesh-Key": "<YOUR_API_KEY>"
        }
      }
    }
  }
}
OpenAI Codex CLI — uses TOML in ~/.codex/config.toml 
[mcp_servers.nefesh]
url = "https://mcp.nefesh.ai/mcp"
Continue.dev — uses YAML in .continue/config.yaml 
mcpServers:
  - name: nefesh
    type: streamable-http
    url: https://mcp.nefesh.ai/mcp

All agents connect via Streamable HTTP — no local installation required.

Tools

ToolDescription
ingest_signalSend raw sensor data. Returns unified stress score + state + adaptation prompt.
get_stateGet current physiological state for a session.
get_historyGet state history over time for a session.
delete_subjectGDPR-compliant deletion of all data for a subject.

Quick test

After adding the config, ask your AI agent:

"What tools do you have from Nefesh?"

It should list the tools above.

State labels

ScoreState
0–19Calm
20–39Relaxed
40–59Focused
60–79Stressed
80–100Acute Stress

Documentation

Privacy

  • No video uploads — edge processing runs client-side
  • No PII stored — strict schema validation
  • GDPR/BIPA compliant — cascading deletion via delete_subject
  • Not a medical device — for contextual AI adaptation only

License

Proprietary. See nefesh.ai/terms.

相關伺服器