Execute a command in an existing Docker sandbox and capture stdout, stderr, exit code, duration, and artifact metadata. Synchronous execution only.
AI agents invoke sandbox_run_job to trigger actions in MedSci Agent. 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.
This tool executes arbitrary commands in a Docker container. Even though it runs in a sandbox, command execution is inherently an Execute category risk because the effects depend entirely on the arguments passed—a malicious agent could run destructive commands (rm -rf, data exfiltration, resource exhaustion) or access sensitive research data.
From the tool's definition Tool name is 'sandbox_run_job' and description states it 'Execute[s] a command in an existing Docker sandbox' with capture of stdout, stderr, and exit code.
Documented attack patterns abuse exactly the kind of access sandbox_run_job gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and MedSci Agent, and nothing reaches the server without passing your rules. This is the rule we recommend for sandbox_run_job:
{
"version": "1",
"default": "deny",
"tools": {
"sandbox_run_job": {
"limits": [
{
"counter": "sandbox_run_job_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}sandbox_run_job 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.
Execute a command in an existing Docker sandbox and capture stdout, stderr, exit code, duration, and artifact metadata. Synchronous execution only. It is categorised as a Execute tool in the MedSci Agent MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the MedSci Agent MCP server in PolicyLayer and add a rule for sandbox_run_job: 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 MedSci Agent. Nothing to install.
sandbox_run_job 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 sandbox_run_job 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 sandbox_run_job. 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.
sandbox_run_job is provided by the MedSci Agent MCP server (omar-a-hassan/medsci-agent). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from MedSci Agent, 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.
28 MedSci Agent tools catalogued and risk-classified — across an index of 43,000+ MCP servers.