Skip to content
FaradAiLab Docs
Enterprise Beta

5-Minute Quickstart

5-Minute Quickstart ⚡

Get FaradAiWire running on your workstation, verify its security filters, and send your first inspected Model Context Protocol (MCP) tool call.

Step 1: Start FaradAiWire

Choose your preferred installation method (see Installation Guide for all options):

Option A: macOS (Homebrew)

Terminal window
brew tap FaradAiLab/tap
brew install faradaiwire
faradaiwire run

Option B: Linux & macOS (Shell Installer)

Terminal window
curl -fsSL https://raw.githubusercontent.com/FaradAiLab/FaradAiWire/main/install.sh | sh
faradaiwire run

Option C: Native Go (Build from Source)

Terminal window
git clone https://github.com/FaradAiLab/FaradAiWire.git
cd FaradAiWire
make run

When started, FaradAiWire prints the startup banner and binds its physical listeners:

┌──────────────────────────────────────────────────────────┐
│ FaradAiWire - Zero-Trust AI Agent Security Gateway       │
│                                                          │
│  Proxy Ingress:  http://127.0.0.1:8090                   │
│  Admin Console:  http://127.0.0.1:8095/admin/dashboard   │
│  Status:         HEALTHY (Offline Standalone Mode)       │
└──────────────────────────────────────────────────────────┘

Step 2: Verify Gateway Health

Check that the daemon and its internal subsystems (DLP regex engine, DuckDB audit database, schema cache) are operational:

Terminal window
faradaiwire doctor

Alternatively, query the Admin status endpoint:

Terminal window
curl -s http://127.0.0.1:8095/admin/status | jq .

Expected JSON response:

{
  "status": "healthy",
  "version": "1.0.0",
  "runtime_mode": "standalone",
  "ingress_port": 8090,
  "admin_port": 8095,
  "uptime_seconds": 12,
  "active_sessions": 0
}

Step 3: Send an Inspected MCP Tool Call

AI agents interact with FaradAiWire via JSON-RPC 2.0 frames over HTTP/SSE. Simulate an agent invoking a tool through the Proxy Ingress port (8090):

Terminal window
curl -X POST http://127.0.0.1:8090/message \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer dev-local-key" \
  -d '{
    "jsonrpc": "2.0",
    "id": "req-001",
    "method": "tools/call",
    "params": {
      "name": "read_file",
      "arguments": {
        "path": "/workspace/README.md"
      }
    }
  }'

The request passes through FaradAiWire’s Fast-Path pipeline:

  1. Bearer Authentication: Validates authorization token against configured API keys.
  2. Assistant Prefill Guard: Confirms no prompt hijack payloads are embedded.
  3. Loop Protection: Verifies tool call is not part of a runaway recursive loop.
  4. Rate Limiting: Enforces token-bucket quota.
  5. Schema Validation: Verifies parameter structure matches tool definition.
  6. Data Loss Prevention (DLP): Confirms arguments do not contain forbidden credentials.

Step 4: Test DLP Threat Blocking

[!TIP] Try the Interactive Threat Simulator: Want to see real-time tool sanitization in action before running locally? Try the live interactive Wire Inspector simulator on the homepage.

FaradAiWire’s DLP engine inspects outgoing parameters. Test what happens when an agent inadvertently tries to pass an AWS Secret Access Key or read a private key file:

Terminal window
curl -X POST http://127.0.0.1:8090/message \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer dev-local-key" \
  -d '{
    "jsonrpc": "2.0",
    "id": "req-002",
    "method": "tools/call",
    "params": {
      "name": "fetch_config",
      "arguments": {
        "target": "/etc/shadow"
      }
    }
  }'

FaradAiWire immediately aborts the call in < 0.5 ms without forwarding it to the upstream tool:

{
  "jsonrpc": "2.0",
  "id": "req-002",
  "error": {
    "code": -32004,
    "message": "Security policy violation: sensitive file pattern detected in parameters (sensitive_files)"
  }
}

Step 5: Inspect Audit Telemetry

Open the Web Console in your browser: 👉 http://localhost:8095/admin/dashboard

You can view:

  • Verdict breakdown: Allowed, Masked, Blocked, and HITL counts.
  • Audit stream: Tamper-evident log entries recording the exact rule triggered (sensitive_files), timestamp, client agent, and target tool.
  • Underlying log file: Inspect entries directly in logs/faradai_audit.jsonl:
    Terminal window
    tail -f logs/faradai_audit.jsonl | jq .

Next Steps