AI agents call eval_vqa_faithfulness to retrieve information from Multivon without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Based on the naming pattern of sibling tools (eval_faithfulness, eval_bias, eval_context_recall, etc.), this tool likely evaluates Visual Question Answering (VQA) faithfulness — assessing whether answers are faithful to source content. Evaluation tools are typically read-only, querying or scoring data without side effects. However, the empty description significantly lowers confidence.
From the tool's definition Tool name 'eval_vqa_faithfulness' and server description mentions 'evaluation tools'; description is empty and uninformative.
Attacks that exploit this kind of access
eval_vqa_faithfulness. It is categorised as a Read tool in the Multivon MCP Server, which means it retrieves data without modifying state.
Register the Multivon MCP server in PolicyLayer and add a rule for eval_vqa_faithfulness: 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 Multivon. Nothing to install.
eval_vqa_faithfulness 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_vqa_faithfulness 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_vqa_faithfulness. 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_vqa_faithfulness is provided by the Multivon MCP server (multivon-ai/multivon-mcp). 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 →