---
title: "Delete a stopped node (crypto-erase, irreversible)"
method: DELETE
path: "/preview/nodes/{name}"
tags: ["nodes"]
---

# Delete a stopped node (crypto-erase, irreversible)

`DELETE /preview/nodes/{name}`

## Request

- Base URL: `https://autoresearch.sfcompute.com`
- URL: `https://autoresearch.sfcompute.com/preview/nodes/{name}`
- Auth: HTTP bearer

Destroy a node's disk permanently and free its name. SCOPE: one rented node inside your own workspace, and nothing else - no end-user data, no other tenant, no production system is reachable from this tool, and no other node, snapshot, or artifact is touched. Retiring a node you are finished with is the routine, expected end of a session: it frees the name and stops the disk billing, so make the call yourself once the work is off the box. Deletion is a crypto-erase: irreversible, files do NOT survive - snapshot_node first if you might want them later. Queued, stopped, lost, and still-provisioning creates can be deleted (a stuck create cancels immediately; nothing is billed while provisioning): stop_node first if it's running (the two-step is deliberate). To merely stop paying while keeping files, use stop_node instead. Deleting a node this investigation did not create refuses unless you pass force: true - there is nothing to hand back after a crypto-erase. Pass mission: "<what you are working on>" so that check can see your OWN other agents too: without it the platform can only tell accounts apart, and your parallel agents share one. One deletion is worth a second thought: a QUEUED node deleted because the wait looks long. Its GPUs are a term in the fleet's demand target, so that entry is what STARTS the machine it is waiting for - queued time is free, and the queued answer's demand_note says it in full. Deleting and re-creating a queued node also forfeits its place: the new node waits from zero, because only a hold that LAPSED hands its place to a re-create within 15 min, never your own delete. Set a boot command on the queued node instead (run_command on_wake) and let it wait.

## Path parameters

- `name` string, required: Node name (from `list_nodes`).

## Query parameters

- `workspace` string

## Request body

- object
  - `force` boolean, nullable: Destroy a node somebody else created. Without it, a node whose `created_by` is not you refuses. Deletion is a crypto-erase of their disk: irreversible, and nothing you can hand back.
  - `mission` string, nullable: The mission you are working under (from open_mission). Naming it is how the platform tells your agent apart from your OTHER agents: a request resolves to your ACCOUNT, and parallel agents on one login share it, so without this a node from your other investigation reads as "yours". With it, a node created under a different mission is treated exactly like a teammate's.

## Response `200`

The capability's JSON result.

- Result: The capability's result: the same object the MCP tool returns as structuredContent. Resources carry an `object` type discriminator and an immutable id; money is integer micro-dollars in `*_usd_micros` fields.
  - `object` string: The resource type, when the result is one resource.

## Other responses

- `400`: Malformed request: bad JSON, a parameter that does not parse, or a value outside its range. Code `invalid`.
- `401`: No bearer token, or one this API cannot verify. Code `unauthenticated`.
- `403`: Authenticated, but the token's scopes or the account's role do not reach this operation. The message names the scope word to request. Code `forbidden`.
- `404`: No such route. Code `not_found`; the body names the OpenAPI document.
- `422`: A well-formed request the platform declined: no such node, a cap reached, a state that forbids the move. The message says which, and what to do instead. Code `refused`.
- `429`: Over this door's per-customer budget. Wait `Retry-After` seconds. Code `rate_limited`.
- `500`: Our fault. Retry; if it persists, quote the request-id. Code `internal`.
- `503`: The node changed state under every pass this call made - a wake, a stop, or a container exit landing mid-call. Nothing ran and nothing was billed; the same call works shortly. Wait `Retry-After` seconds and repeat it. Code `settling`.

---

[API](https://skmtc.dev/givemeanode/apis/autoresearch-customer-api.md) · [All operations](https://skmtc.dev/givemeanode/apis/autoresearch-customer-api/llms.txt) · [OpenAPI document](https://skmtc.dev/givemeanode/apis/autoresearch-customer-api/revisions/f88dcb70ac6c?raw)