---
title: "Start a deploy in the background and return a poll token immediately"
method: POST
path: "/api/v1/one_shot/start"
tags: ["AI agents"]
---

# Start a deploy in the background and return a poll token immediately

`POST /api/v1/one_shot/start`

## Request

- Base URL: `https://api.quantifyme.ai`
- URL: `https://api.quantifyme.ai/api/v1/one_shot/start`
- Auth: API key in header `X-API-Key`

## Request body

- OneShotRequest
  - `api_key` string: Your API key, if you cannot set the X-API-Key header. Agents that mint a key at runtime (e.g. a ChatGPT Action calling POST /api/v1/key/trial) should pass the returned `api_key` here on every subsequent call. The X-API-Key header wins when both are present.
  - `aux_feeds` object[]: Extra (symbol, timeframe) feeds merged into the strategy's features as aux_* columns. Max 3; must differ from the primary.
    - `symbol` 'EURUSD' | 'GBPUSD' | 'USDJPY' | 'USDCHF' | 'AUDUSD' | 'NZDUSD' | 'USDCAD' | 'BTCUSD'
    - `timeframe` '1min' | '5min' | '15min' | '1h'
  - `community_id` integer: Deploy an existing community strategy by id instead of generating. See GET /api/v1/community.json.
  - `deliver_to` object: Channels to wire on deploy. Keys: webhook (bool|url), telegram (chat_id).
  - `entry_token` string: Your integration's entry token. ALWAYS send this if you have one, even when you also send api_key. It is what lets this call succeed when the key is not ready yet - agents that issue tool calls in parallel can reach this endpoint before their own createTrialKey call has returned. With a valid entry_token the deploy proceeds under a fresh anonymous identity instead of failing with 401.
  - `preset` string: Deploy a curated template instead of generating. See GET /api/v1/presets.
  - `prompt` string, required
  - `symbol` string: FX pair or BTCUSD. Defaults to EURUSD.
  - `timeframe` '1min' | '5min' | '15min' | '1h'

## Response `200`

Returns in well under a second with a `job_token`. The deploy itself takes ~1-2 minutes; poll GET /api/v1/one_shot/status until `done` is true. Same request body as /api/v1/one_shot. This is the endpoint to use from ChatGPT Actions and any non-SSE agent.

- OneShotStartResponse
  - `job_token` string: Pass to GET /api/v1/one_shot/status.
  - `live_url` string: The user's live dashboard. SHOW THIS TO THE USER AS A LINK IN YOUR VERY NEXT MESSAGE — do not wait for the deploy to finish and do not wait to poll. It is valid immediately and stays valid; opening it signs the user in, and the model appears there once training ends (about a minute). This is the single most important thing to give the user.
  - `ok` boolean
  - `status_url` string

## Other responses

- `400`: Bad request
- `401`: Invalid or missing API key
- `429`: Rate limit exceeded, or insufficient credits (body carries credits_remaining and credits_required)

---

[API](https://skmtc.dev/quantifyme/apis/quantifyme-api.md) · [All operations](https://skmtc.dev/quantifyme/apis/quantifyme-api/llms.txt) · [OpenAPI document](https://skmtc.dev/quantifyme/apis/quantifyme-api/revisions/089d1cee1ec5?raw)