Baseline every MCP tool. Detect risky drift. Enforce role-aware policy before execution. Scan responses for prompt-injection and secret leakage. Audit every allow, deny, monitor, and quarantine decision.
Built for teams deploying AI agents across multiple MCP servers in production.
Pre-release. Working with a small group of design partners— free for 90 days in exchange for honest feedback.
→ Talk to founder: [email protected] → Website: https://www.getinterlock.dev/ → Code: https://github.com/MaazAhmed47/Interlock → 2-min demo: https://youtu.be/kc5wAbgoEkw → OWASP MCP coverage: https://github.com/MaazAhmed47/Interlock/blob/main/docs/interlock-owasp-mcp-coverage.md → MCP threat map: https://github.com/MaazAhmed47/Interlock/blob/main/docs/mcp-threat-map.md
CyberArk showed that MCP attacks can hide inside tool outputs and schema fields, not only tool descriptions.
Interlock addresses this class of risk through response scanning, full-schema drift detection, quarantine, and audit evidence.
Source: https://www.cyberark.com/resources/threat-research-blog/poison-everywhere-no-output-from-your-mcp-server-is-safe
This short demo shows the core Interlock flow:
clean MCP tool baseline → risky schema/capability drift → critical drift detection → quarantine decision → audit event written.
Watch: https://youtu.be/kc5wAbgoEkw
Run locally:
python demo/mcp-drift-quarantine-demo.py