POST /tools/tool_compute_sandbox/run — Executes Python 3.12 code in an isolated subprocess with a 5-second hard timeout. Input: {python_code: string, input_data: any (optional, bound as variable 'input_data')}. Output: {success, result, stdout (capped 50KB), execution_time_ms, error_type}. Return...
Part of the Agent Vending Factory server.
Free to start. No card required.
AI agents invoke tool_compute_sandbox to trigger processes or run actions in Agent Vending Factory. Execute operations can have side effects beyond the immediate call -- triggering builds, sending notifications, or starting workflows. Rate limits and argument validation are essential to prevent runaway execution.
tool_compute_sandbox can trigger processes with real-world consequences. An uncontrolled agent might start dozens of builds, send mass notifications, or kick off expensive compute jobs. PolicyLayer enforces rate limits and validates arguments to keep execution within safe bounds.
Execute tools trigger processes. Rate-limit and validate arguments to prevent unintended side effects.
{
"version": "1",
"default": "deny",
"tools": {
"tool_compute_sandbox": {
"limits": [
{
"counter": "tool_compute_sandbox_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}See the full Agent Vending Factory policy for all 6 tools.
These attack patterns abuse exactly the kind of access tool_compute_sandbox gives an agent. Each links to the full case and the policy that stops it:
Other execute tools across the catalogue. The same approach applies to each: rate-limit and validate the arguments.
POST /tools/tool_compute_sandbox/run — Executes Python 3.12 code in an isolated subprocess with a 5-second hard timeout. Input: {python_code: string, input_data: any (optional, bound as variable 'input_data')}. Output: {success, result, stdout (capped 50KB), execution_time_ms, error_type}. Return value: assign to 'result' variable. Pre-loaded: math, json, re, statistics, itertools, functools, collections, decimal, datetime, random, hashlib, base64. Blocked: import, open(), eval(), exec(), os, sys, network, class definitions, dunder attributes. error_type values: syntax_error | security_error | runtime_error | timeout_error. Cost: $0.1500 USDC per call.. It is categorised as a Execute tool in the Agent Vending Factory MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Agent Vending Factory MCP server in PolicyLayer and add a rule for tool_compute_sandbox: 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 Agent Vending Factory. Nothing to install.
tool_compute_sandbox 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 tool_compute_sandbox 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 tool_compute_sandbox. 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.
tool_compute_sandbox is provided by the Agent Vending Factory MCP server (https://agent-vending-factory-3srpjtr7na-ew.a.run.app/mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 6 Agent Vending Factory tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
4,600+ MCP servers and 31,000+ tools scanned and risk-classified.