Fetch evidence for low-confidence claims using uncertainty-guided retrieval. For each claim, the inner LLM generates smart search queries (adversarial by default — designed to DISPROVE the claim), executes web searches, and evaluates each result
AI agents call arsr_retrieve_evidence to retrieve information from Mcp Arsr without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
arsr_retrieve_evidence is fundamentally a data retrieval tool that fetches and evaluates search results to inform claim verification. While it executes web searches, this is a standard read-only query operation, not code execution or destructive action. The severity is low because search queries are inherently safe and reversible, with no ability to modify data, execute arbitrary code, or cause financial harm.
From the tool's definition Tool performs 'Fetch evidence' and 'executes web searches' — these are query and retrieval operations with no side effects.
Attacks that exploit this kind of access
Fetch evidence for low-confidence claims using uncertainty-guided retrieval. For each claim, the inner LLM generates smart search queries (adversarial by default — designed to DISPROVE the claim), executes web searches, and evaluates each result. It is categorised as a Read tool in the Mcp Arsr MCP Server, which means it retrieves data without modifying state.
Register the Mcp Arsr MCP server in PolicyLayer and add a rule for arsr_retrieve_evidence: 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 Arsr. Nothing to install.
arsr_retrieve_evidence 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 arsr_retrieve_evidence 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 arsr_retrieve_evidence. 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.
arsr_retrieve_evidence is provided by the Mcp Arsr MCP server (jayarrowz/mcp-arsr). 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 →