AI agents use create_project to create or update resources in MCP Files — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your MCP Files environment.
This tool creates (writes) a new project directory structure. The operation is reversible—the sibling `remove_directory` tool can undo it. While it requires valid credentials ('API key allows project management'), the blast radius is limited to creating project artifacts within a constrained filesystem workspace, not executing arbitrary commands, deleting critical data, or moving money.
From the tool's definition Tool description explicitly states it 'Create[s] a new `.projects/<uuid>` directory', which is a reversible data creation operation.
Attacks that exploit this kind of access
Create a new .projects/<uuid> directory when the API key allows project management. Requires a valid UUID. It is categorised as a Write tool in the MCP Files MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the MCP Files MCP server in PolicyLayer and add a rule for create_project: 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 Files. Nothing to install.
create_project 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 create_project 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 create_project. 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.
create_project is provided by the MCP Files MCP server (tspspi/mcpfiles). 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 →