AI agents call ai_classify to retrieve information from Onion without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool analyzes and categorizes text input to provide classification results. This is purely a read operation—it examines data and returns insights without side effects, state changes, or irreversible actions. No data is created, modified, deleted, or executed. It falls squarely within the Read category.
From the tool's definition Tool description indicates it performs text classification tasks: '对文本进行分类，支持情感分析、主题分类、意图识别，或自定义分类标签' (classifies text, supports sentiment analysis, topic classification, intent recognition, or custom classification labels).
Attacks that exploit this kind of access
对文本进行分类，支持情感分析、主题分类、意图识别，或自定义分类标签。. It is categorised as a Read tool in the Onion MCP Server, which means it retrieves data without modifying state.
Register the Onion MCP server in PolicyLayer and add a rule for ai_classify: 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 Onion. Nothing to install.
ai_classify 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 ai_classify 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 ai_classify. 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.
ai_classify is provided by the Onion MCP server (onion-ai/mcp-server). 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 →