AI agents call extract_info to retrieve information from MCP_3 without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool performs a read-only search operation across paper directories to retrieve information. It has no side effects, does not modify data, execute code, delete content, or involve financial transactions. The search capability aligns with the 'Read' category for data retrieval operations. Severity is low because misuse would only expose existing paper metadata without enabling destructive or system-level attacks.
From the tool's definition Tool name 'extract_info' and description '在所有话题目录中搜索特定论文的信息' (search for information of specific papers in all topic directories) indicate a search/query operation that retrieves paper information without modification or deletion.
Attacks that exploit this kind of access
在所有话题目录中搜索特定论文的信息。. It is categorised as a Read tool in the MCP_3 MCP Server, which means it retrieves data without modifying state.
Register the MCP_3 MCP server in PolicyLayer and add a rule for extract_info: 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 MCP_3. Nothing to install.
extract_info 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 extract_info 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 extract_info. 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.
extract_info is provided by the MCP_3 MCP server (sunhuiowo/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 →