Finding security vulnerabilities is easy. Remediation is where engineering velocity dies. Traditionally, security consultancies email a 60-page PDF report. Developers must manually parse vague descriptions, try to replicate findings, research fixes, and schedule engineering sprints.
By supporting the Model Context Protocol (MCP), Lorikeet Security bridges the gap between offensive findings and automated engineering remediation. With our MCP Server (ptaas/mcp/), developer tools like Claude Code, Claude Desktop, and Cursor connect directly to your live penetration test data.
The 60-Second Patch Loop: An engineer opens Claude Code in their terminal, prompts: "Fetch open Critical findings from Lorikeet Security on our billing API and patch the IDOR flaw in billing_controller.py." Claude pulls the live reproduction payload, refactors the source code, and writes a regression test in seconds.
Connecting Claude Code & Desktop to the Lory MCP Server
The Lorikeet Security MCP endpoint is located at https://lorikeetsecurity.com/talon/mcp. To connect your IDE or Claude client, configure your claude_desktop_config.json:
Granular Scopes & Enterprise Audit Logging
Security is built into the protocol. Tokens are minted with least-privilege scopes:
findings:read: Read-only access to active findings, CVSS scores, and reproduction evidence.kb:read: Access to verified remediation patterns and framework-specific patch guidance.compliance:read: Verification of control status against SOC 2, ISO 27001, and PCI DSS frameworks.
Every tool call executed by an AI assistant is cryptographically logged in the mcp_audit_log table, ensuring complete accountability for compliance auditors.
Supercharge Your Developer Remediation with Lory MCP
Fix vulnerabilities before auditors or adversaries find them. Lorikeet Security combines autonomous offensive testing with native Model Context Protocol integration.
