MCP Security Checklist
The practitioner’s security framework for production MCP deployments.
MCP adoption is outpacing the security guidance for it — teams ship servers with no auth, no validation, and no monitoring. The checklist gives engineering and security teams a structured, actionable framework: seven domains, a prioritised Top 10, and four formats.
- Status
- Active
- Domains
- 7 security domains
- Formats
- Markdown · JSON · YAML · Web
- License
- MIT
What it covers
A shared vocabulary and control set so no MCP security review starts from scratch.
Top 10 critical controls
mTLS, least-privilege tool scoping, input-schema validation, behavioural anomaly detection.
7 security domains
Authentication through to executive risk framing — every domain maps to actionable controls, not guidelines.
CI/CD integration
JSON and YAML formats run checklist validation as a gate in your deployment pipeline.
Four formats
Markdown for engineers, JSON for automation, YAML for pipelines, an interactive web version for SOC teams.
The seven domains
Each contains actionable controls with implementation guidance.
mTLS enforcement, token scoping, least-privilege tool access.
Schema enforcement, injection detection, sanitisation patterns.
Minimal surface area, permission boundaries, tool inventory.
Session lifecycle, token rotation, replay prevention.
Behavioural anomaly detection, audit trails, alerting.
Egress controls, network segmentation, TLS configuration.
Business risk framing, compliance mapping, escalation guidance.
More from Helixar Labs
All projectsVerifiable human delegation for agentic AI.
HDP extended to physical AI: robots, vehicles, surgical systems.
Scan, harden, sign and attest build artifacts before they ship.