Remove a git worktree sandbox by sandboxId.
AI agents call sandbox_delete to permanently remove resources in Promethean OS MCP — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.
This tool permanently deletes a git worktree sandbox, which is an irreversible operation. Once deleted, the sandbox and any uncommitted work within it cannot be recovered. This qualifies as Destructive rather than Write because deletion is not reversible.
From the tool's definition Tool name is 'sandbox_delete' and description states 'Remove a git worktree sandbox by sandboxId' — the verb 'Remove' indicates irreversible deletion of a git worktree, which cannot be undone and results in loss of the sandbox environment.
Attacks that exploit this kind of access
Remove a git worktree sandbox by sandboxId. It is categorised as a Destructive tool in the Promethean OS MCP MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.
Register the Promethean OS MCP server in PolicyLayer and add a rule for sandbox_delete: 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 Promethean OS MCP. Nothing to install.
sandbox_delete is a Destructive tool with critical risk. Critical-risk tools should be blocked by default and only enabled with explicit human approval.
Yes. Add a rate_limit block to the sandbox_delete 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_delete. 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_delete is provided by the Promethean OS MCP server (octave-commons/mcp). 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 →