Rerank a set of papers by quality signals. Takes DOIs or paper IDs from search results,
AI agents call rerank_papers to retrieve information from Paper Search without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Reranking is a read/compute operation that takes existing paper identifiers and returns them in a new order based on quality signals. No data is created, modified, deleted, or any external system triggered. The blast radius of misuse is minimal — at worst an AI agent gets a suboptimal ordering of papers.
From the tool's definition 'Rerank a set of papers by quality signals. Takes DOIs or paper IDs from search results' — this operation reorders/scores existing data without modifying or deleting anything
Attacks that exploit this kind of access
Rerank a set of papers by quality signals. Takes DOIs or paper IDs from search results,. It is categorised as a Read tool in the Paper Search MCP Server, which means it retrieves data without modifying state.
Register the Paper Search MCP server in PolicyLayer and add a rule for rerank_papers: 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 Paper Search. Nothing to install.
rerank_papers 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 rerank_papers 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 rerank_papers. 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.
rerank_papers is provided by the Paper Search MCP server (upascal/paper-search-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 →