accept_character_voice_line
ChatGPTAccept or undo extra time for a replacement voice line. Read the timing changes in inspect_and_resume_character_voices first. The final accepted line can publish the replacement voices.
add_blank_card
ChatGPTAdd a blank Card at the chosen position using the same editor action. Keep requestId after a lost reply. Use the stored script version from read_project when required.
add_voice_line
ChatGPTGenerate and add one new spoken line from quote_add_voice_line after confirmation of its exact text, credit price and revision. Keep requestId and the exact input after a lost response; pending or uncertain work must not be started with a new ID.
animate_coverage
ChatGPTAnimate reviewed existing Shots after customer confirmation. Some tiers split long dialogue into shorter Shots. Copy acceptedQuotes from quote_coverage_animation. Keep requestId after a lost reply; a completed request returns the original partial or complete receipt.
archive_card
ChatGPTArchive a Card from the saved video. Supply the observed active Card order and revisions. The Card can be restored.
archive_shot
ChatGPTArchive a Shot timeline item from a Card, with the requested ripple behavior and observed Card revision. Keep requestId after a lost response.
attach_element
ChatGPTAttach an accessible library character, location or object. Repeating the same source keeps one attachment. No media is generated.
cancel_character_voice_change
ChatGPTCancel a pending voice replacement. This does not undo work already accepted by the generation service.
change_character_voice
ChatGPTReplace one speaker's generated voice using an accepted quote. Requires customer confirmation. Keep the same request ID after a lost reply; inspect_and_resume_character_voices gives progress.
copy_video_script_to_draft
ChatGPTSave the supplied video script as a separate editable draft. Keeps the original video intact. Use requestId to recover a lost reply.
create_card
ChatGPTSave a Card with a complete customer script in a project scene. Use visual action and @Narrator: "spoken words". No media generation. Keep idempotencyKey and content after a lost reply.
create_character
ChatGPTCreate a reusable character from its name, description, appearance, and clothing, with an optional personality (at most 600 characters) and speech, how they talk (at most 400). Your character limit applies. This does not generate an image. Keep the same request ID after a lost reply.
create_characters_from_image
ChatGPTCreate reusable characters from an image attached in ChatGPT. Uses the same image analysis, moderation and character limit as the character library. Keep requestId after a lost reply.
create_element
ChatGPTCreate a reusable location or object from its customer description and attach it. Saves text only. Keep requestId after a lost reply.
create_project
ChatGPTCreate an empty editable video project. Saves settings and its first scene without generating media. Keep requestId after a lost reply.
detach_element
ChatGPTDetach a project element where the editor permits it. Existing references can prevent removal.
draft_shot
ChatGPTSave a complete manual Shot draft, script, opening frame, motion and references in a Card. This does not generate media. Action Shots require 4–15 seconds and prepared character/location references; object references and character speech are not supported for Action Shots. Read the current Card revision first.
edit_shot_image
ChatGPTEdit or reframe a Shot image using customer instructions and project references. A completed take replaces its active image selection; previous takes remain available. Requires native approval at the quoted image price. Keep requestId after a lost reply; a new take needs a new ID and approval. Animate the resulting image with a separate video request.
edit_shot_script
ChatGPTEdit a current Short Shot before it has keyframe or segment records. Generated Shot scripts are read-only. For a manual draft use revise_shot. Keep requestId after a lost response.
edit_timeline
ChatGPTApply one timeline gesture from the embedded LongStories editor at the exact Card revision. Model callers use the individually named timeline tools.
edit_voice_line
ChatGPTGenerate replacement speech for one existing line from quote_edit_voice_line after confirmation of its exact text, credit price and revision. Replaces saved speech, without rewriting the generated Shot script or making lip-sync media. Keep requestId and the exact input after a lost response.
generate_card_coverage
ChatGPTGenerate the reviewed Card coverage after customer confirmation. Replaces that Card's current Shots. Copy the exact quote fields from quote_card_coverage. Keep requestId after a lost reply; a completed request returns its original receipt.
generate_shot
ChatGPTGenerate a reviewed manual Shot. Uses native user confirmation of the exact current price and revision. Keep the original draft idempotencyKey. A repeat observes the same paid attempt; it cannot purchase a retry.
generate_shot_sound
ChatGPTGenerate or remove a Shot's overlay sound effects through the existing editor action. This action currently charges zero customer credits. Native confirmation is required. Keep requestId after a lost response.
generate_video
ChatGPTCreate and generate a complete video through the same LongStories creation flow as the website. Supports ideas, full scripts, and owned uploaded narration or songs. Use audioId from list_media_library for uploaded audio. Show the current credit price and get the user's approval before generation. Keep the same operationKey and input after a lost reply; a new key means a new paid video. Returns a project and generation run to observe and edit.
import_shot_image
ChatGPTApply an existing completed image from your account to a Shot. sceneId is the Shot ID used by the image picker. sourceAssetId must be an owned media asset; URLs and storage keys are not accepted.
insert_timeline_hold
ChatGPTInsert a hold of a positive duration in milliseconds at a zero-based Shot-lane position, or at its end. This repacks the lane and changes the timing of later items. Keep the same holdId when retrying. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
inspect_and_resume_character_voices
ChatGPTFind the video's speakers and their IDs for quote_character_voice, including before the first voice change. Return current change status and lines that need a timing decision. This also resumes accepted work and can publish a ready voice replacement or end an expired review.
inspect_export
ChatGPTCheck whether current project content is ready to export. Reports output dimensions, duration and a content signature. Does not start rendering.
inspect_project
ChatGPTInspect saved project state with separate Card, Shot, and editor-run pages. Read creation.runId with observe_video_creation to resume a Create run. Use read_project for the full script, read_card for details, and read_export to confirm the saved video includes current edits. This does not start or retry work.
link_voice_line_to_shot
ChatGPTLink an existing voice-line timeline item to a visual item in the same Card. Does not generate speech or lip-sync media. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
list_media_library
ChatGPTList owned image assets or uploaded audio. Use image asset IDs for references/import and creationReady audio IDs for narration or song creation. If audio is not creationReady, upload it with upload_audio first. Private storage keys are not returned.
list_projects
ChatGPTList this account's saved projects. Use the returned project ID to read or edit a video. Page through all results when needed.
list_scripts
ChatGPTList saved scripts owned by this account, including video scripts and separate drafts.
merge_cards
ChatGPTMerge a Card with the specified next Card using the editor's media and timeline rules. Keep requestId after a lost response.
move_timeline_item
ChatGPTMove one Card timeline item to a local start time in milliseconds. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
observe_export
ChatGPTRead a prior export attempt. This never starts another render. A successful attempt can still be behind current edits; read_export checks that.
observe_run
ChatGPTRead the saved status of an editor generation run. Queued or running work is not a reason to resubmit. Read the Card and takes after success.
observe_video_creation
ChatGPTRead the saved state and completed output of a full video generation in your account. Use the projectId and runId returned by generate_video. Generation continues on LongStories while you are away. An accepted request does not mean the video is complete.
open_editor
ChatGPTOpen the Video editor inside ChatGPT. With no project ID, show this account's projects. With a project ID, open its saved Cards and Shots. The editor uses the same product actions as LongStories. Opening it does not generate or change anything.
prepare_element
ChatGPTGenerate an attached element reference after native confirmation of its current revision and quoted credits. Repeated requests use the same saved attempt. Read elements until ready.
prepare_next_episode
ChatGPTPrepare the next episode from an owned completed video, just as Create does. Creates no series: a video in no series gets seriesId null, and generate_video makes the series with the episode. Fills missing story records of an existing series; no video is generated and no customer credits are charged. Returns safe story records and creationSettings. Add the user's next-episode idea and an operationKey, then quote_video_creation and generate_video. Read the source project's Cards if any story record is missing. Keep requestId after a lost reply.
quote_add_voice_line
ChatGPTQuote a new spoken line for one Card. Returns its exact change, credit price and Card revision without generating speech or charging credits. Pass that quote unchanged to add_voice_line.
quote_card_coverage
ChatGPTRead the fixed price for a Card storyboard or animated coverage. Returns the saved Card revision and the Shots that would be replaced. Does not plan, generate, or spend credits.
quote_character_voice
ChatGPTQuote the credits and current video revision for replacing one speaker's generated voice across the video.
quote_character_voice_retry
ChatGPTQuote a new generation of a replacement voice line with revised text. Use this when the replacement line does not fit.
quote_coverage_animation
ChatGPTRead the video price for existing Shots, including any split needed by the chosen tier. Returns exact per-Shot prices and revisions. Does not generate or spend credits.
quote_edit_voice_line
ChatGPTQuote a replacement for one existing spoken line in a Card. Returns its exact change, credit price and Card revision without generating speech or charging credits. Pass that quote unchanged to edit_voice_line.
quote_shot_image
ChatGPTRead the current price and effective model for a Shot image edit, angle change, or guidance regeneration. Guidance requires mode=guidance. This creates no media.
quote_shot_video
ChatGPTRead the current video price for one Shot, including its model, duration, and lip-sync choice. This does not generate video or modify settings.
quote_video_creation
ChatGPTReview the price and settings for a complete LongStories video, using the same creation flow as the website. Start from an idea, a full LongStories script, or narration or song audio already uploaded to your account. Use audioId from list_media_library for narration or songs. This does not start video generation. Keep the same operationKey for the matching generation request.
quote_voice_line
ChatGPTQuote the voice-line form in the embedded LongStories editor. Model callers use quote_add_voice_line or quote_edit_voice_line.
read_archive
ChatGPTRead archived Cards and Shot timeline items with the identities and revisions needed to restore them.
read_card
ChatGPTRead one Card's full script, Shot scripts, timing, speech, media and editing revisions.
read_character_library
ChatGPTRead your saved characters, their descriptions, appearance, clothing, and selected voices. Returns at most 24 characters per page.
read_creation_options
ChatGPTList this account's universes, visual styles and supported creation settings. Read these before creating a project.
read_elements
ChatGPTRead project character, location and object handles and reference preparation status. Optionally list accessible library elements. Use these handles in scripts and references.
read_export
ChatGPTRead the saved final video and compare it with current edits. Only saved.state=current confirms the export includes those edits.
read_media_request
ChatGPTRead the accepted run for an exact earlier image or video request after a lost reply. Supply the same request ID and complete original input. This never starts or retries work. A not_found result only means an accepted run is not visible yet; it does not prove that no work started.
read_operation
ChatGPTRead an accepted customer operation after a lost reply, including saved result IDs. Pending or uncertain work must not be submitted with a new request ID. Generation runs and exports also have their own observation tools.
read_project
ChatGPTRead the full stored script and current Card-derived script, scenes, Cards, Shots and customer settings. Use current revisions before edits. Internal generation instructions are never returned.
read_script
ChatGPTRead the full text and revision of an owned saved script. Video scripts are read-only here; current video edits use Cards and Shots.
read_script_format
ChatGPTRead the public LongStories script syntax and a complete valid example. Use validate_script before quoting a full script for video creation.
read_shot
ChatGPTRead a Shot's current revision and recover its original manual draft operation key and price review. Use this to continue a saved draft in a new chat.
read_shot_generation
ChatGPTRead the accepted action for the exact original generate_shot input after a lost reply. This never starts, retries, or charges generation. A not_found result can still mean the original request is pending. Keep its draft key and review.
read_shot_preview
ChatGPTRead only the selected completed image and video for an owned Shot. Does not return take history or private generation instructions.
read_shot_takes
ChatGPTList saved image or video takes for a Shot, with safe preview links and selection status.
read_voice_choices
ChatGPTFind selectable voices by name, language, accent, or description. Returns preview links and at most 24 voices per page.
regenerate_shot_image
ChatGPTGenerate a new Shot image from the customer's visual description and optional project location. A completed take replaces its active image selection; previous takes remain available. Quote mode=guidance first. Requires native approval. Keep requestId after a lost reply. A deliberate new take needs a new ID and approval.
regenerate_shot_video
ChatGPTAnimate the Shot's selected image or make a new video take, with optional motion and lip-sync edits. A completed take can replace the selected clip in the saved Card; previous takes remain available. Requires native confirmation of the current quote. Keep requestId after a lost reply; a deliberate new take needs a new ID and approval. Read the saved run and takes for progress.
remove_shot_lipsync
ChatGPTRemove the saved lip-sync selection from one visual item in a Card. Does not generate new media. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
remove_timeline_item
ChatGPTRemove one item from a Card timeline and ripple later items. This changes the saved video. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
remove_voice_line
ChatGPTRemove an individual voice line. If it is linked to lip sync, explicitly confirm the break. Keep requestId after a lost response.
remove_watermark
ChatGPTRemove LongStories watermark and outro branding from the owned project settings. Requires the owner's existing watermark-removal plan entitlement and current project revision. Preserves custom intro assets. Read the project and export it again to include this change in the final video.
rename_card
ChatGPTRename a Card at its current revision.
reorder_card
ChatGPTMove a Card to a one-based position, optionally in another project Scene. Keep requestId after a lost response and read changed Card revisions.
reorder_shot_lane_item
ChatGPTMove a Shot-lane item to a new zero-based lane position within its Card. The lane includes visuals and holds. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
reorder_shot_visual
ChatGPTMove a Shot visual to a new one-based visual order within its Card. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
request_voice_line
ChatGPTSubmit the quoted voice-line form from the embedded LongStories editor after exact text, price and revision confirmation. Keep the request ID after a lost response. Model callers use add_voice_line or edit_voice_line.
resize_timeline_item
ChatGPTResize one Card timeline item in milliseconds. Speech timing is protected and later items ripple with the change. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
restore_card
ChatGPTRestore an archived Card. Supply the current active Card order and revisions.
restore_shot
ChatGPTRestore an archived Shot timeline item through the editor's Card rules. Keep requestId after a lost response.
restore_shot_sound
ChatGPTRestore the Shot's previous saved sound effects. Does not generate audio. Keep requestId after a lost response.
retry_character_voice_line
ChatGPTGenerate a replacement voice line again with revised text and an accepted credit quote. Requires customer confirmation. Keep the same request ID after a lost reply.
review_shot
ChatGPTRead an ungenerated manual Shot draft and its current credit quote. Use its original idempotencyKey.
revise_shot
ChatGPTRevise an ungenerated manual Shot's script, opening frame, motion and references together. Pass the current Shot revision. Timing and template remain as drafted. Accepted media attempts cannot be changed with this tool.
save_script_draft
ChatGPTSave a separate script draft at its exact revision. This does not change an existing video's Cards, Shots or media.
select_shot_take
ChatGPTSelect an existing image or video take for a Shot. No media generation. The shared editor checks the take belongs to this Shot and is ready.
set_card_captions
ChatGPTSet captions for one Card timeline item. Pass enabled, text, or both, and its current Card revision.
set_card_clip_audio
ChatGPTChoose the original clip audio or the project's dubbed voice for one visual item.
set_card_duration
ChatGPTSet the saved duration of one Card in milliseconds, subject to its content and speech limits. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
set_cards_audio_mix
ChatGPTChange SFX mix, voiceover mute or clip audio across selected Cards. Supply each Card revision. Read all Cards after a partial or uncertain result.
set_character_voice
ChatGPTSet a saved character's voice for future generation. Set voiceId to null for Auto. Existing generated speech is unchanged; use change_character_voice to replace it.
set_project_clip_audio
ChatGPTSet the project's clip audio default and update selected Cards. Pass project and Card revisions. Read the project after a partial result.
set_script_archived
ChatGPTArchive or restore a separate script draft. Video scripts cannot be archived through this action.
set_shot_sfx_mix
ChatGPTSet mute state, volume from zero to one, or both for a visual item's sound-effects mix within a Card. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
set_timeline_item_audio
ChatGPTSet mute state, volume from zero to one, or both for one audio-bearing Card timeline item. A visual clip supports volume changes; use set_card_clip_audio to select its audio source. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
split_card
ChatGPTSplit a Card at a millisecond position inside it. Keep requestId after a lost response; do not repeat under a new ID without reading the deck.
start_export
ChatGPTRender the saved project with a supported quality and export preset. Keep attemptId after a lost reply. New attemptId means a new render. Check readiness first.
trim_shot_source
ChatGPTTrim a visual item's selected video to a source start and end in milliseconds. The saved video selection stays the same; later timeline items ripple with the trim. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
trim_voiceover_silence
ChatGPTTrim the saved silence around a voiceover item using its recorded speech timing. Does not regenerate audio.
unlink_voice_line
ChatGPTRemove an existing voice-line timeline item's visual link within a Card. The voice line remains in the Card. Supply the exact current Card revision. Read the Card again after a conflict or an uncertain result.
update_project_settings
ChatGPTChange the project title, captions, transition or voiceover playback enablement. This does not generate media. A title change leaves scripts, timing and media unchanged. Read the saved project after a conflict.
upload_audio
ChatGPTUpload narration or a song attached in ChatGPT for video creation. Uses the same audio format, duration and plan checks as the Create page. Returns an owned audioId for generation. Keep requestId after a lost reply.
upload_shot_image
ChatGPTUpload an image attached in ChatGPT and select it for an owned Shot. The normal image moderation and file limits apply. This does not generate a new image. Keep requestId after a lost reply.
validate_script
ChatGPTValidate customer script syntax without generation or model calls. Returns exact issues to fix. Creation still checks settings, plan and storage rules.