Dry-run placing one tile and solve without modifying the draft
AI agents call test_placement to retrieve information from Ice Puzzle without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a simulation or query tool that tests behavior without altering state. It retrieves information about what would happen if a tile were placed, making it a read-only operation despite its computational nature. The explicit mention of 'without modifying the draft' confirms no data is permanently altered.
From the tool's definition The tool performs a 'dry-run' that does not modify the draft, meaning it executes a test operation to observe outcomes (solve) without side effects or persistent changes.
Attacks that exploit this kind of access
Dry-run placing one tile and solve without modifying the draft. It is categorised as a Read tool in the Ice Puzzle MCP Server, which means it retrieves data without modifying state.
Register the Ice Puzzle MCP server in PolicyLayer and add a rule for test_placement: 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 Ice Puzzle. Nothing to install.
test_placement 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 test_placement 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 test_placement. 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.
test_placement is provided by the Ice Puzzle MCP server (wmoten/ice-puzzle-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 →