MCP Tool Risk Explorer
MCP and agentic AI systems become more powerful as they gain access to tools, resources, prompts, and external systems. This explorer shows how risk increases as an AI system moves from answering questions to reading data, taking actions, and modifying real-world systems.
Source alignment: This explorer is inspired by public MCP security guidance and Andrew's work as Principal Co-author of the CIS Controls v8.1 MCP Companion Guide. It is simplified for education and portfolio demonstration and is not an official CIS product or assessment.
How is risk scored?
Each tool capability carries a risk weight by tier, low-impact reads count least, destructive or high-privilege capabilities count most:
- Low-impact 2 · Moderate 5 · High-impact 10 · Critical 16
Selected capabilities are summed, and missing controls add to the total. The score maps to four bands:
- Low 0–9 · Moderate 10–22 · High 23–42 · Critical 43+
It is a simplified educational model for building intuition about how tool access drives risk, not a calibrated quantitative assessment.
A secure MCP deployment enforces controls at the tool boundary, before the agent can affect external systems.
Step 1, Select Capabilities
Select every tool, resource, or capability this AI system can access. Risk updates in the panel on the right.
Read-only access to public or pre-approved content. Minimal autonomous impact on external systems.
Access to internal or proprietary data sources. Retrieval may expose sensitive organizational information.
Write access or actions that affect external systems, people, or records. Mistakes can be costly or hard to reverse.
System-level access or irreversible actions with broad organizational impact. Require the strongest controls.
Step 2, Controls In Place
Check every control that is currently in place for this system. Unchecked controls that are relevant to your selected capabilities will contribute to the risk score.
Labels (all / mod+ / high+ / crit) show which capability tiers activate each control gap.
Select capabilities on the left to generate your risk profile.
Sources & Inspiration
- CIS Controls v8.1 MCP Companion Guide (Principal Co-author)
- CIS Controls v8.1 AI and LLM Companion Guide (Collaborator)
- CIS Controls v8.1 AI Agents Companion Guide (Collaborator)