Convert a chat message into one or more board-task drafts.
Generates a list of `{ subject, description }` task drafts from a chat message. In the Nurama web app this backs the "Generate with Nu" option in a message's create-task menu; the returned `tasks` array seeds the task creator's draft list. **Compound capability requirement.** Unlike Polish, this endpoint requires BOTH the `boards` capability (checked against the project) and the `ai` capability (checked against the workspace); both `workspaceId` and `projectId` must belong to an active subscription. Either capability missing returns `capabilityNotAvailable`. **Context window.** `contextMessages` is the preceding chat thread (oldest first). The server caps at 10 messages and each entry's `content` at 8000 characters; supply the messages immediately before the focal one so the model can resolve pronouns and references. **Permission.** Requires `canAiGenerateTasks` (held by all workspace/project admins and by the creator, reviewer and reviewerAdmin roles). **Pre-prompt.** If the workspace or project has `aiTaskGenerationPreprompt` configured, it is appended to the system prompt as an override block (project value wins per-key). Separate from `aiCustomPreprompt` so admins can tune task style without affecting Polish or Chat output. Caps: - `messageText` is capped at 8000 characters. - `contextMessages` is capped at 10 entries; each entry's `content` is capped at 8000 characters and `authorName` at 100. - At most 10 tasks are returned.
Generates a list of { subject, description } task drafts from a
chat message. In the Nurama web app this backs the "Generate with Nu"
option in a message's create-task menu; the returned tasks array
seeds the task creator's draft list.
Compound capability requirement. Unlike Polish, this endpoint
requires BOTH the boards capability (checked against the project)
and the ai capability (checked against the workspace); both
workspaceId and projectId must belong to an active subscription.
Either capability missing returns capabilityNotAvailable.
Context window. contextMessages is the preceding chat
thread (oldest first). The server caps at 10 messages and each
entry's content at 8000 characters; supply the messages immediately
before the focal one so the model can resolve pronouns and references.
Permission. Requires canAiGenerateTasks (held by all
workspace/project admins and by the creator, reviewer and
reviewerAdmin roles).
Pre-prompt. If the workspace or project has
aiTaskGenerationPreprompt configured, it is appended to the
system prompt as an override block (project value wins per-key).
Separate from aiCustomPreprompt so admins can tune task style
without affecting Polish or Chat output.
Caps:
messageTextis capped at 8000 characters.contextMessagesis capped at 10 entries; each entry'scontentis capped at 8000 characters andauthorNameat 100.- At most 10 tasks are returned.
Authorization
bearerAuth In: header
Request Body
application/json
TypeScript Definitions
Use the request body type in TypeScript.
Response Body
application/json
application/json
application/json
application/json
curl -X POST "https://example.com/ai/task-generation" \ -H "Content-Type: application/json" \ -d '{ "workspaceId": "ef0efa32-d1c1-43d4-a5e2-fe7b4f00403c", "projectId": "5a8591dd-4039-49df-9202-96385ba3eff8", "messageText": "string" }'{ "tasks": [ { "subject": "Draft launch checklist", "description": "List every step needed before go-live." }, { "subject": "Notify stakeholders", "description": "Email the launch plan to leadership." } ], "billedCredits": 4, "balanceAfter": 9996, "eventId": "0193e4a1-b2c3-7000-8000-00000000abcd"}