AI agents call top_matches to retrieve information from Crosswalk without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool fetches and returns ranked job matches based on stored fit scores. Even the optional scoring of unscored jobs appears to be a read/compute operation that queries and caches results rather than modifying any user-facing data. No creation, deletion, or financial action is involved.
From the tool's definition 'Best-fit jobs for a stored resume, from cached fit scores' — retrieves and ranks cached scoring data; 'Optionally scores recent unscored jobs first' is a read/compute operation with no persistent side effects described.
Attacks that exploit this kind of access
Best-fit jobs for a stored resume, from cached fit scores. Optionally scores recent unscored jobs first (one model call each). It is categorised as a Read tool in the Crosswalk MCP Server, which means it retrieves data without modifying state.
Register the Crosswalk MCP server in PolicyLayer and add a rule for top_matches: 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 Crosswalk. Nothing to install.
top_matches 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 top_matches 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 top_matches. 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.
top_matches is provided by the Crosswalk MCP server (mohakgarg5/crosswalk-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 →