Codex CLI guide

You can add the Snyk MCP server to Codex CLI to secure code generated with agentic workflows through an LLM. This can be achieved in several ways. When you use it for the first time, the MCP server will ask for trust and trigger authentication if necessary.

Prerequisites

• Install the Codex CLI

• Install the Snyk CLI

• Install the Snyk MCP

Install Codex CLI

Consult the official OpenAI Codex CLI documentation on GitHub for Codex CLI installation instructions.

Install the Snyk MCP Server in Codex CLI using npx

This installation instruction assumes you have a Node.js local development environment setup with the npx executable.

To install the Snyk MCP Server using Codex's own .codex/config.toml file, proceed with the following instructions:

• Create or edit the .codex/config.toml file

• Insert the following mcp_servers configuration block to the file:

  Copy

  [mcp_servers.snyk-security]
  command = "npx"
  args = ["-y", "snyk@latest", "mcp", "-t", "stdio"]

• Restart the Codex CLI

Install the Snyk MCP Server in Codex CLI using Snyk CLI

This installation instruction assumes you have installed the Snyk CLI and the snyk executable is available in your system's path.

To install the Snyk MCP Server using the .codex/config.toml Codex file, follow these instructions:

• Create or edit the .codex/config.toml file

• Insert the following mcpServers configuration block to the file:

  Copy

  [mcp_servers.snyk-security]
  command = "/absolute/path/to/snyk"
  args = ["mcp", "-t", "stdio"]

• Restart the Codex CLI

Setting up the Snyk MCP Server

As a one-time setup, you may need to authenticate and trust the current Project directory. If required, the agentic workflow will likely manage this automatically.

These workflows are mostly automatically executed by the underlying model and the agentic code assistant, and you will need to approve them using a browser confirmation dialog. The process should look similar to this:

You can explicitly prompt to authenticate your account as follows:

Authenticate my Snyk account

If you are not authenticated, this prompt opens a browser for you to authenticate to the Snyk Web UI, and then you can continue with the agent.

Examples

Scanning for security vulnerabilities

In the chat, you can tag specific files or functions and ask the agent to scan your code and third-party dependencies of the Project.

Scan this directory for code security & dependency vulnerabilities and security issues

Codex CLI will invoke Snyk scanning, provide output and insights based on the results that Snyk returned:

"Secure at inception" rules

To practice Secure at Inception, Snyk recommends adopting rules that align the LLM for secure code generation and workflows.

The following is a suggested instruction for Codex CLI's AGENTS.md context file:

Always run Snyk Code scanning tool for new first party code generated.
Always run Snyk SCA scanning tool for new dependencies or dependency updates.
If any security issues are found based on newly introduced or modified code or dependencies, attempt to fix the issues using the results context from Snyk.
Rescan the code after fixing the issues to ensure that the issues were fixed and that there are no newly introduced issues.
Repeat this process until no issues are found.

Add the following contents to the AGENTS.md file in your root directory of your Project:

# Project security best practices

- Always run Snyk Code scanning tool for new first party code generated.
- Always run Snyk SCA scanning tool for new dependencies or dependency updates.
- If any security issues are found based on newly introduced or modified code or dependencies, attempt to fix the issues using the results context from Snyk.
- Rescan the code after fixing the issues to ensure that the issues were fixed and that there are no newly introduced issues.
- Repeat this process until no issues are found.