AI agents use save_scene to create or update resources in Godot MCP — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Godot MCP environment.
This tool modifies and persists data (scene files) in a way that is typical of Write operations. While the changes are saved to disk, they are not destructive because scene edits can be undone, reverted, or re-edited. The severity is medium because unintended scene modifications could corrupt game assets or disrupt project state, but this is bounded to a single scene and recoverable.
From the tool's definition Tool description states 'Save changes to a scene file' — this creates or modifies scene data persistently within a Godot project, but changes are reversible (scenes can be re-edited or reverted via version control).
Attacks that exploit this kind of access
Save changes to a scene file. It is categorised as a Write tool in the Godot MCP MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Godot MCP server in PolicyLayer and add a rule for save_scene: 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 Godot MCP. Nothing to install.
save_scene is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the save_scene 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 save_scene. 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.
save_scene is provided by the Godot MCP server (jamesdowzard/godot-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 →