SnozeDevelopers
API referenceAi

Run an AI generator and return its structured result

Runs the named generator as a single non-streaming model call with a schema-constrained output, allowing one repair attempt when the model misses the schema, and returns the decoded result with token usage. Requires a workspace member with the `ai:write` permission, and the call consumes the workspace AI allowance: a workspace that has reached its limit is refused before the model is called. Send an `Idempotency-Key` header to make retries safe; a replay within the window returns the original response.

POST
/v1/workspaces/{workspaceId}/ai/generate

Authorization

apiKey
headerAuthorizationBearer <snoze_sk_…>

Workspace API key created in Workspace settings → API keys. Scopes on the key bound what it can read and write.

Path Parameters

workspaceId*string

Request Body

application/json
  1. body
generator*string
input*unknown
options?|null

Response Body

Success

application/json
  1. response
data*
Example Request
curl -X POST 'https://api.snoze.dev/v1/workspaces/string/ai/generate' \  -H 'Content-Type: application/json' \  -d '{  "generator": "string",  "input": null}'
Example Response
{  "data": {    "runId": "string",    "generator": "string",    "output": null,    "diagnostics": [      {        "severity": "info",        "code": "string",        "message": "string",        "path": "string"      }    ],    "usage": {      "inputTokens": 0,      "outputTokens": 0,      "cachedInputTokens": 0,      "cacheWriteInputTokens": 0,      "model": "string"    }  }}

List the AI generators available in a workspace

Returns the descriptor for every registered one-shot generator that the generate operation can run. Requires an authenticated caller who is a member of the workspace.

List the operations this workspace always approves

Returns the operation names someone has marked "always allow" from an approval card, which is what stops the agent asking before it runs them. Requires a workspace member with the `ai:read` permission.