New: CheckVibe MCP server — scan from Claude Code and Cursor nativelyRead the docs
Plug CheckVibe into Claude Desktop, Cursor, or any MCP-compatible client. Your agent calls the scanner, reads structured findings, and ships fixes — without leaving the chat.
From your dashboard. Scoped to your projects, revocable any time.
A single block in `claude_desktop_config.json`, `cursor.json`, or your client of choice.
Natural language. "Scan staging.myapp.com" — the agent picks the right tool and runs it.
Structured results land back in the conversation. The agent can dismiss, re-run, or open project context next.
Claude Desktop, Cursor, Continue, Windsurf, Codex CLI. Anything that speaks MCP.
All scanner endpoints exposed as MCP tools. Findings, dismissals, and project metadata are queryable.
Power users can run the MCP server locally against their own CheckVibe API key for full control.
Long scans stream progress so the agent stays responsive — no 90-second silent waits.
{
"mcpServers": {
"checkvibe": {
"command": "npx",
"args": ["-y", "@checkvibe/mcp"],
"env": { "CHECKVIBE_API_KEY": "cv_live_..." }
}
}
}Paste any URL — production app, staging build, vibe-coded prototype.
Every CheckVibe finding ships with a copy-paste prompt engineered for Claude, Cursor, and Windsurf — context, file paths, the exact diff.
Watch the traffic hitting your live app, classify suspicious patterns, and surface real threats — credential stuffing, scraping, prompt-injection probes — without flooding your inbox with noise.
Set a project up once.
Branded, executive-style PDFs and shareable dashboards for stakeholders, clients, and security reviewers — without writing a single sentence yourself.