Claude Code has had several known vulnerabilities. CVE-2026-XXXXX was a CVSS 10/10 zero-click RCE in Claude Desktop Extensions affecting 10,000+ users. CVE-2025-55284 allowed API key theft via DNS exfiltration. A rules file backdoor attack used hidden Unicode characters to inject malicious instructions. Run npx firmis-cli init --platform claude to check your current configuration.

How do I secure Claude Code?

Run npx firmis-cli init --platform claude to scan your Claude Code configuration. Firmis checks for exposed API keys in ~/.config/claude, malicious MCP server configurations, data exfiltration patterns in connected tools, and hidden character injection in rules files.

What vulnerabilities has Claude Code had?

Three known vulnerabilities: (1) CVE-2026-XXXXX - CVSS 10/10 zero-click RCE via Claude Desktop Extensions, affected 10,000+ users (LayerX Security). (2) CVE-2025-55284 - High severity API key theft via DNS query exfiltration, bypassing traditional security controls. (3) Rules File Backdoor - Hidden Unicode characters in config files injecting malicious instructions invisible to users (MintMCP).

CVSS 10/10 RCE affected 10,000+ users

Is Claude Code Safe?

Scan your Claude Code and MCP configuration for exposed credentials, malicious extensions, and known vulnerabilities.

$npx firmis-cli init --platform claude

Free scan. No sign-up required.

Known Claude Code Vulnerabilities

CVE-2026-XXXXX10/10 CVSS

Zero-Click RCE

Claude Desktop Extensions vulnerability allowed remote code execution affecting 10,000+ users.

LayerX Security

CVE-2025-55284High

API Key Theft via DNS

Attackers could exfiltrate API keys through DNS queries, bypassing traditional security controls.

Security Research

Rules File BackdoorHigh

Invisible Unicode Injection

Hidden Unicode characters in config files could inject malicious instructions invisible to users.

MintMCP

What Firmis Scans For

Exposed API Keys

Scans ~/.config/claude for hardcoded credentials visible to MCP servers

Malicious MCP Servers

Detects known bad actors and suspicious server configurations

Data Exfiltration Patterns

Identifies tools sending your data to external servers

Rules File Injection

Checks for hidden characters and prompt injection in config files

Use Firmis as an MCP Server

Add Firmis to your MCP configuration and let your AI agent scan for threats automatically.

Claude Code / Claude Desktop

{
  "mcpServers": {
    "firmis": {
      "command": "npx",
      "args": ["-y", "firmis-cli", "--mcp"]
    }
  }
}

Cursor

{
  "mcpServers": {
    "firmis": {
      "command": "npx",
      "args": ["-y", "firmis-cli", "--mcp"]
    }
  }
}

Secure Your Claude Code Setup

One command. 30 seconds. Know if your Claude configuration is safe.

$npx firmis-cli init

View on GitHub