AI agents call get_friends to retrieve information from Splitwise without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and displays friend information without modifying, deleting, or executing any operations. It is a lookup/query function that has no destructive or financial impact. The blast radius of misuse is minimal—an AI agent could only read friend data, not modify finances or relationships.
From the tool's definition Tool name is 'get_friends' and description states 'List all Splitwise friends' — a retrieval operation with no side effects. The description explicitly indicates it queries and returns data (IDs, names, emails, balances) for informational purposes only.
Attacks that exploit this kind of access
List all Splitwise friends with their IDs, names, emails, and current balances. Use this to look up user IDs before creating expenses. It is categorised as a Read tool in the Splitwise MCP Server, which means it retrieves data without modifying state.
Register the Splitwise MCP server in PolicyLayer and add a rule for get_friends: 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 Splitwise. Nothing to install.
get_friends 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 get_friends 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 get_friends. 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.
get_friends is provided by the Splitwise MCP server (saharshpatel24/splitwise-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 →