Skip to main content

Connect Synli to your AI assistant

MCP is included in Pro and Enterprise. It uses your organization's API access and shared scan quota.

The client exposes only six tools. Scan starts use approved saved targets, and scans consume the same quota as other Synli scans. A report link is private and expires.

Client compatibility

  • Codex: all six tools tested, including scan start and PDF report links with a separate human confirmation.
  • Claude Desktop: target listing, scan status, summaries and findings tested. Scan start and report links are not supported in the tested version because it does not present Synli's confirmation form.
  • GitHub Copilot in VS Code: compatibility test pending. Do not assume full support from MCP protocol support alone.

Before you start

Use Node.js 24.22.2 or a newer 24.x release. Tested secure storage: Windows x64, macOS 15 arm64 and Ubuntu 24.04 x64 with Secret Service. Create and approve a saved scan target in Synli.

1. Install the local client

npm install --global https://synli.ai/downloads/synli-mcp-server-0.1.0.tgz
Download checksum

2. Create an API key

In Synli, open API keys and create a short-lived key with targets:read, scans:read, scans:write and reports:read. Run this command in your own terminal and paste the key only into its hidden prompt. Never paste it into an AI chat or client configuration.

Open API keys
synli-mcp auth login --origin https://api.synli.ai --profile synli_customer_mcp

3. Connect your assistant

Use the following server command and arguments in your client's MCP settings. The profile name identifies the key in your operating system's secure store; it is not the key itself.

Run npm root -g to find the installation directory. Replace the first argument below with that directory followed by /@synli/mcp-server/dist/cli-bin.js. Use forward slashes in Windows paths too.

{
  "command": "node",
  "args": [
    "/absolute/path/to/node_modules/@synli/mcp-server/dist/cli-bin.js",
    "serve",
    "--origin",
    "https://api.synli.ai",
    "--profile",
    "synli_customer_mcp"
  ]
}

Disconnect

Revoke the key in Synli and remove the local credential with the command below.

synli-mcp auth logout --origin https://api.synli.ai --profile synli_customer_mcp