AI agents use pause_project to create or update resources in Vercel MCP Server — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Vercel MCP Server environment.
Pausing a project is a reversible write action that modifies the project's state from active to paused. While it has significant operational impact (taking the project offline), it is not irreversible — the project can be unpaused/resumed. This makes it Write rather than Destructive. Severity is high because misuse would take a production project offline.
From the tool's definition 'Pause a project' — pauses a running Vercel project, which stops deployments/serving but is a reversible state change (project can be resumed).
Documented attack patterns abuse exactly the kind of access pause_project gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Vercel MCP Server, and nothing reaches the server without passing your rules. This is the rule we recommend for pause_project:
{
"version": "1",
"default": "deny",
"tools": {
"pause_project": {
"limits": [
{
"counter": "pause_project_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
} pause_project stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Pause a project. It is categorised as a Write tool in the Vercel MCP Server MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Vercel MCP Server MCP server in PolicyLayer and add a rule for pause_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 Vercel MCP Server. Nothing to install.
pause_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 pause_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 pause_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.
pause_project is provided by the Vercel MCP Server MCP server (quegenx/vercel-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 154 Vercel MCP Server tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
154 Vercel MCP Server tools catalogued and risk-classified — across an index of 42,500+ MCP servers.