Checkmarx application security, built for AI coding agents. This repository holds two ways to bring Checkmarx One into an AI-assisted workflow: an MCP server your assistant can call, and a Claude Code plugin that scans code as it is written.
Use them together or separately — they solve different halves of the problem.
A hosted MCP server that connects any MCP-capable AI client — Claude, Cursor, Copilot, Windsurf, Kiro — to Checkmarx One. It exposes scanning, findings, project management, and AI-generated remediation as tools your assistant can call in conversation.
Reach for this when you want your assistant to scan projects, investigate findings, or fix them on request — in whichever AI client you already use. Configure it once (see examples/ for per-client config) and ask.
For more details → README-MCP.md
→ plugins/cx-devassist/README.md
A fail-closed security gate for Claude Code. Before Claude writes a
file, runs a command, or calls a tool, the Checkmarx cx CLI scans the proposed action. Real
vulnerabilities are blocked rather than silently allowed — and so is the case where the scanner
itself can't be trusted to run. Findings are remediated through the bundled Checkmarx MCP server.
Reach for this when you want the check to be automatic and non-optional rather than something someone remembers to ask for.
For more details → plugins/cx-devassist/README.md
- docs/usage.md — the MCP tool catalog and example workflows
- docs/authentication.md — API key and OAuth2 setup
- docs/troubleshooting.md — connection, auth, and scan issues
Apache 2.0 — see LICENSE for details. It governs everything in this repository, including the
cx-devassist plugin.
See CONTRIBUTING.md for guidelines, and SECURITY.md to report a vulnerability.
Website: Checkmarx.
© 2026 Checkmarx Ltd.