Add security hardening and bias-t support

Security improvements:
- Add interface name validation to prevent command injection
- Fix XSS vulnerability in pager message display
- Add security headers (X-Content-Type-Options, X-Frame-Options, etc.)
- Disable Werkzeug debug PIN
- Add security documentation

Features:
- Add bias-t power support for SDR dongles across all modes

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

docs/SECURITY.md

@@ -0,0 +1,89 @@
+# Security Considerations
+
+INTERCEPT is designed as a **local signal intelligence tool** for personal use on trusted networks. This document outlines security considerations and best practices.
+
+## Network Binding
+
+By default, INTERCEPT binds to `0.0.0.0:5050`, making it accessible from any network interface. This is convenient for accessing the web UI from other devices on your local network, but has security implications:
+
+### Recommendations
+
+1. **Firewall Rules**: If you don't need remote access, configure your firewall to block external access to port 5050:
+   ```bash
+   # Linux (iptables)
+   sudo iptables -A INPUT -p tcp --dport 5050 -s 127.0.0.1 -j ACCEPT
+   sudo iptables -A INPUT -p tcp --dport 5050 -j DROP
+
+   # macOS (pf)
+   echo "block in on en0 proto tcp from any to any port 5050" | sudo pfctl -ef -
+   ```
+
+2. **Bind to Localhost**: For local-only access, set the host environment variable:
+   ```bash
+   export INTERCEPT_HOST=127.0.0.1
+   python intercept.py
+   ```
+
+3. **Trusted Networks Only**: Only run INTERCEPT on networks you trust. The application has no authentication mechanism.
+
+## Authentication
+
+INTERCEPT does **not** include authentication. This is by design for ease of use as a personal tool. If you need to expose INTERCEPT to untrusted networks:
+
+1. Use a reverse proxy (nginx, Caddy) with authentication
+2. Use a VPN to access your home network
+3. Use SSH port forwarding: `ssh -L 5050:localhost:5050 your-server`
+
+## Security Headers
+
+INTERCEPT includes the following security headers on all responses:
+
+| Header | Value | Purpose |
+|--------|-------|---------|
+| `X-Content-Type-Options` | `nosniff` | Prevent MIME type sniffing |
+| `X-Frame-Options` | `SAMEORIGIN` | Prevent clickjacking |
+| `X-XSS-Protection` | `1; mode=block` | Enable browser XSS filter |
+| `Referrer-Policy` | `strict-origin-when-cross-origin` | Control referrer information |
+| `Permissions-Policy` | `geolocation=(self), microphone=()` | Restrict browser features |
+
+## Input Validation
+
+All user inputs are validated before use:
+
+- **Network interface names**: Validated against strict regex pattern
+- **Bluetooth interface names**: Must match `hciX` format
+- **MAC addresses**: Validated format
+- **Frequencies**: Validated range and format
+- **File paths**: Protected against directory traversal
+- **HTML output**: All user-provided content is escaped
+
+## Subprocess Execution
+
+INTERCEPT executes external tools (rtl_fm, airodump-ng, etc.) via subprocess. Security measures:
+
+- **No shell execution**: All subprocess calls use list arguments, not shell strings
+- **Input validation**: All user-provided arguments are validated before use
+- **Process isolation**: Each tool runs in its own process with limited permissions
+
+## Debug Mode
+
+Debug mode is **disabled by default**. If enabled via `INTERCEPT_DEBUG=true`:
+
+- The Werkzeug debugger PIN is disabled (not needed for local tool)
+- Additional logging is enabled
+- Stack traces are shown on errors
+
+**Never run in debug mode on untrusted networks.**
+
+## Reporting Security Issues
+
+If you discover a security vulnerability, please report it by:
+
+1. Opening a GitHub issue (for non-sensitive issues)
+2. Emailing the maintainer directly (for sensitive issues)
+
+Please include:
+- Description of the vulnerability
+- Steps to reproduce
+- Potential impact
+- Suggested fix (if any)