Nodaro DocsPipelines
Start a Story to Video pipeline from a prompt, watch its stages, approve or refine them in chat, and branch a finished pipeline from any stage with MCP tools.
Available on Nodaro Cloud
The pipeline tools drive the Story to Video pipeline from an AI assistant. One prompt starts a pipeline that writes a script, creates the cast and places, plans and renders the shots, and merges a finished film. The assistant can watch the stages, approve them, refine them in a chat, and branch a finished pipeline from any stage. The canvas version of the same engine is the Story to Video node.
The pipeline tools exist only on Nodaro Cloud. They appear only when you granted the pipelines:read, pipelines:execute and pipelines:approve permissions.
start_pipeline, branch_pipeline, chat_pipeline_stage and apply_chat_proposal start or continue a pipeline's work. They are not listed where the deployment does not offer the film pipeline, the availability id film-pipeline in its nodes.allow and nodes.deny lists. See Editions and profiles. A call that still arrives there is refused with 403 node_not_available before anything is reserved. The read tools stay listed, so you can still follow a pipeline made before.
The stages
A pipeline runs these stages in order: script, characters, objects, locations, shot_list, scene_images, animate_audio_edit and post_merge. In auto mode it runs them all on its own, and stops only when a planned match cut between two shots breaks. The run then waits until every break is accepted over the REST API. No MCP tool accepts a break. In manual and guided mode it stops at approval gates, and the stage waits with the status awaiting_approval.
start_pipeline
Starts a new Story to Video pipeline from a story prompt.
Permission: pipelines:execute. Credits: the credits of every generation the pipeline runs. get_pipeline_status shows the credits as it goes.
| Parameter | Type | Notes |
|---|---|---|
story_prompt | string | Required. What the film is about, up to 4,000 characters. |
format | string | trailer, short_film, music_video, reel or commercial. Default reel. |
target_duration_seconds | integer | The length of the film, from 5 to 600 seconds. Default 15. |
mode | string | auto (default) runs to the end unattended, except at a match-cut break. manual and guided stop at approval gates. |
output_resolution | string | 480p, 720p, 1080p or 4K. Default 720p. |
music_enabled | boolean | Adds music. Default true. |
narration_enabled | boolean | Adds narration. Default false. |
lipsync_enabled | boolean | Syncs the characters' lips to their lines. Default false. |
Returns: the new pipeline id. Poll get_pipeline_status, and pipeline_pending_approvals while a stage waits.
Narration and lip sync are off by default when an assistant starts a pipeline, so an unattended run does not spend credits on them unless you ask for them.
get_pipeline_status
Returns the state of a pipeline: its status (queued, running, awaiting_approval, completed, failed, cancelled or forked), the current stage, the credit counters, the mode, and failure_reason when it failed.
Permission: pipelines:read. Credits: free.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. The pipeline id. |
pipeline_pending_approvals
Lists the stages that wait for approval, each with a snapshot of its output. An empty list means nothing waits: the run is in auto mode with no match-cut break, or it is still generating.
Permission: pipelines:read. Credits: free.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. The pipeline id. |
chat_pipeline_stage
Sends a message to the refinement director about a stage that waits for approval, for example "make the ending darker". The pipeline must be in guided mode.
Permission: pipelines:approve.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. The pipeline id. |
stage | string | Required. The stage that waits: script or post_merge. |
message | string | Required. Your message, up to 8,000 characters. |
Returns: the director's reply and, when it proposes an edit, a proposed_change. Apply the change with apply_chat_proposal.
apply_chat_proposal
Accepts the change a director reply proposed, and approves the stage with that change.
Permission: pipelines:approve. Credits: free.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. The pipeline id. |
stage | string | Required. The stage of the chat. |
turn_id | string | Required. The id of the director's reply whose change you accept. |
get_pipeline_stage_chat
Lists the messages of a stage's chat, oldest first. The list is empty when nobody has written yet.
Permission: pipelines:read. Credits: free.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. The pipeline id. |
stage | string | Required. The stage of the chat. |
branch_pipeline
Creates a new pipeline that re-runs a completed pipeline from one stage. The stages before it, and their characters, objects and locations, are copied as approved. The chosen stage and every later stage run again.
Permission: pipelines:execute. Credits: the credits of the stages that run again.
| Parameter | Type | Notes |
|---|---|---|
pipeline_id | string | Required. A pipeline with the status completed. |
from_stage | string | Required. script, characters, objects, locations, shot_list, scene_images, animate_audio_edit or post_merge. |
Returns: the new pipeline id.
Frequently asked questions
What does a Story to Video pipeline produce?
A finished film from one story prompt. The pipeline writes the script, creates the characters, objects and locations, plans the shots, generates the scene images, animates them with audio, and merges the final cut.
What is the difference between auto, manual and guided mode?
Auto runs from the prompt to the finished film on its own. It stops only when a planned match cut between two shots breaks, until every break is accepted over the REST API. No MCP tool accepts a break. Manual and guided stop at approval gates. Guided also lets you refine the script and the final cut in a chat before you approve.
Can I re-run a pipeline from the middle?
Yes, once it is completed. branch_pipeline copies the stages before the one you choose as approved, and runs that stage and every later stage again as a new pipeline.
Why are the pipeline tools missing from my assistant?
They need the pipelines permissions, and they exist only on Nodaro Cloud. Reconnect and allow the pipeline permissions on the consent screen.
Related
Story → Video
MCP tools reference
Pipelines
Film Director
Last updated on
Apps and components
Find published Nodaro apps and components, read their inputs, run them and retry failed runs from an AI assistant with list_apps, run_app and retry_app_run.
Prompts and text
Turn a rough idea into a tested prompt, write a scene-by-scene video script, and let an LLM judge pick the best of several results with Nodaro MCP tools.