A/B test two sets of AI outputs against the same rubric. Returns score distributions, a winner declaration, and statistical significance for each criterion.
AI agents call eval_compare to retrieve information from Scan Your Ai Toolkit without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
eval_compare performs comparative analysis of AI outputs against a rubric, returning computed metrics and comparisons. This is fundamentally a data retrieval and analysis operation with no side effects on the system, no code execution capability, no data modification, and no irreversible actions. It belongs in the Read category with low severity due to its limited blast radius.
From the tool's definition Tool description indicates it 'Returns score distributions, a winner declaration, and statistical significance' — operations that retrieve and analyze data without modifying, deleting, or executing external actions.
Attacks that exploit this kind of access
A/B test two sets of AI outputs against the same rubric. Returns score distributions, a winner declaration, and statistical significance for each criterion. It is categorised as a Read tool in the Scan Your Ai Toolkit MCP Server, which means it retrieves data without modifying state.
Register the Scan Your Ai Toolkit MCP server in PolicyLayer and add a rule for eval_compare: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Scan Your Ai Toolkit. Nothing to install.
eval_compare is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the eval_compare rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for eval_compare. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
eval_compare is provided by the Scan Your Ai Toolkit MCP server (sakthivelchan89/scan_your_ai_toolkit). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →