Start a Scrapybara Ubuntu instance. Use it as a desktop sandbox to access the web or run code. Always present the stream URL to the user afterwards so they can watch the instance in real time.
AI agents invoke start_instance to trigger actions in Scrapybara MCP. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
Starting an instance that can run arbitrary code and bash commands is an Execute-category action. While not immediately destructive, it has high blast radius because an AI agent can use the instance to run arbitrary commands, access external systems, modify files, and perform side effects whose consequences depend on what code is executed.
From the tool's definition Tool description states it 'Start[s] a Scrapybara Ubuntu instance' for users to 'access the web or run code', and the server description mentions the ability to 'control instances through mouse/keyboard actions and bash commands'.
Documented attack patterns abuse exactly the kind of access start_instance gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Scrapybara MCP, and nothing reaches the server without passing your rules. This is the rule we recommend for start_instance:
{
"version": "1",
"default": "deny",
"tools": {
"start_instance": {
"limits": [
{
"counter": "start_instance_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}start_instance stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Start a Scrapybara Ubuntu instance. Use it as a desktop sandbox to access the web or run code. Always present the stream URL to the user afterwards so they can watch the instance in real time. It is categorised as a Execute tool in the Scrapybara MCP MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Scrapybara MCP server in PolicyLayer and add a rule for start_instance: 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 Scrapybara MCP. Nothing to install.
start_instance is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the start_instance 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 start_instance. 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.
start_instance is provided by the Scrapybara MCP server (scrapybara/scrapybara-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Scrapybara MCP, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
5 Scrapybara MCP tools catalogued and risk-classified — across an index of 43,000+ MCP servers.