add_dialogue_alternatives
ChatGPTSaves one or more ALTERNATE VERSIONS of a dialogue line, next to the version in use (in the app: the "Alt" button on a dialogue line). Additive: the line in the screenplay does not change and no existing alternative is removed; the writer picks a version in the app, or you can switch it with choose_dialogue_alternative. Find the line with get_scene (each block has an index). Requires the Owner or Editor role.
add_outline_beat
ChatGPTAppends a beat to the outline (additive: it never touches existing beats). From the outline the user can then generate the scenes in the app. Requires the Owner or Editor role.
add_research_note
ChatGPTSaves a research note in the project (it lands in the app's Research Inbox).
add_scene
ChatGPTAppends a scene to the end of the project. heading = scene heading (e.g. "INT. HOUSE — DAY"), body = text in screenplay format. Requires the Owner or Editor role.
add_studio_shots
ChatGPTCreates one or more shots in the project's AI Studio (appended to the existing ones, in the given order). Each shot can come from a screenplay scene (sceneNumber: title and source text are taken from the scene) or be free-standing (explicit title/sourceText); imagePrompt and videoPrompt are optional and can be written later with set_shot_prompts. ADDITIVE: it never touches existing shots. Requires the Owner or Editor role.
choose_dialogue_alternative
ChatGPTPuts one of the saved alternate versions of a dialogue line in use: the chosen version becomes the line in the screenplay and the previous line takes its place among the alternatives (a swap, nothing is lost). get_scene shows each dialogue line with its alternatives. Requires the Owner or Editor role.
create_branch
ChatGPTCreates a "what if" branch from a scene: an alternate version of the story (e.g. an ending B) that runs in parallel without touching the original. The scene it creates starts the branch; add_scene calls right after continue it. heading and body (optional) fill the first scene of the branch. The user compares branches in the app with the timeline selector and the Canvas board. Requires the Owner or Editor role.
create_project
ChatGPTCreates a new empty project (screenplay) for the user.
get_credits
ChatGPTAI credit balance of the connected account (MCP tools never spend credits).
get_documents
ChatGPTReads the project's other documents: subject, treatment and outline (the beats, with the isConverted flag for those already turned into a scene).
get_project
ChatGPTReads a project: title and full scenes (number, heading, text, logline). Scenes with an empty logline don't have one yet.
get_scene
ChatGPTReads ONE scene by number (12A works too), with typed blocks (scene-heading, action, character, parenthetical, dialogue). More precise and lighter than get_project when you need to rewrite.
get_studio_cast
ChatGPTReads the cards in the AI Studio Cast & Location tab: characters, locations and free REFERENCES (products, objects, styles — kind prop/style/other) with id, name, description and whether they already have an image. Cards with an image are reusable references on shots; to SEE them use view_cast_image. To create or fill them use upsert_studio_cast; the image itself is generated from the app (it spends credits).
get_studio_shots
ChatGPTReads the project's AI Studio shots (screenplay → image → video pipeline): id, title, source scene, source text and prompts. Empty prompts still have to be written (set_shot_prompts); to create new shots use add_studio_shots.
insert_scene
ChatGPTInserts a new scene right AFTER an existing one, in the middle of the story (add_scene only appends at the end). heading = scene heading (e.g. "INT. HOUSE — DAY"), body = text in screenplay format. The scenes that follow shift by one; when the project's numbering is locked for production, the new scene gets a letter number (e.g. 12A) and the others keep theirs. Requires the Owner or Editor role.
list_characters
ChatGPTLists the project's characters, derived from the character blocks: name, number of lines and scenes they appear in. Sorted by line count.
list_projects
ChatGPTLists the user's ShotClick projects (owned and shared), with id and title.
list_research_notes
ChatGPTReads the project's research notes (Inbox and notes attached to scenes), with source and date. Most recent first, max 200.
rename_project
ChatGPTChanges the title of a project (the screenplay's title, also shown on the title page). Requires the Owner or Editor role.
search_project
ChatGPTSearches text across the project's scenes (case-insensitive). Returns scene, block type and an excerpt for each match (max 40).
set_loglines
ChatGPTSets the loglines (1-2 sentence summaries) of one or more scenes in a single call, found by number. Scenes without a logline show an empty logline field in get_project. An empty logline ("") clears the existing one. Requires the Owner or Editor role.
set_shot_prompts
ChatGPTWrites the prompts (image and/or video) of one or more AI Studio shots in a single call, found by id (from get_studio_shots). Omitted fields stay unchanged; an empty string clears them. Requires the Owner or Editor role.
set_title_page
ChatGPTFills in the screenplay's title page: author, "based on" credit, contact details and date (the title itself is the project title: see rename_project). Only the fields passed are changed; pass "" to clear one. The cover image is not affected. get_documents returns the current title page. Requires the Owner or Editor role.
update_block
ChatGPTRewrites ONE line (block) of a scene and leaves the rest of the scene untouched: the line keeps its identity, so comments pinned to it, revision marks and alternate versions stay attached. Prefer it to update_scene for small edits. On a dialogue line, keepPreviousAsAlternative=true saves the old text among the alternate versions instead of discarding it. Find the line with get_scene (each block has an index). Requires the Owner or Editor role.
update_document
ChatGPTWrites the project's subject or treatment. SAFETY: if the document is not empty it refuses, unless overwrite=true is passed explicitly — so a client can't wipe out work by mistake. Requires the Owner or Editor role.
update_scene
ChatGPTRewrites the content of an existing scene, found by number. Keeps its id, number and position. Requires the Owner or Editor role.
upsert_studio_cast
ChatGPTCreates or fills cards in the AI Studio Cast & Location tab. Each entry: with assetId (from get_studio_cast) it UPDATES the name/description of the existing card (omitted fields stay unchanged); without assetId it CREATES a new card (kind + name required: "character", "location", or a free REFERENCE — "prop" for products and objects, "style" for visual styles, "other"). The description doubles as the prompt the app uses to generate the reference image — write it visual and concrete (characters: age, face, build, clothes; locations: setting, period, light, materials; products: shape, materials, packaging, background). It never touches images or other cards. Requires the Owner or Editor role.
view_cast_image
ChatGPTShows the generated image of a Cast & Location card in AI Studio (character, location or free reference: product, object, style), resized. assetId comes from get_studio_cast. Use it to actually see faces, environments and references before writing prompts that mention them.
view_shot_image
ChatGPTShows the generated image (keyframe) of an AI Studio shot, resized: use it to see what the shot actually contains before writing or refining the video prompt with set_shot_prompts, so the motion you describe matches the image. take: 1-based index among the shot's takes (from get_studio_shots), omitted = active take. Videos cannot be displayed this way.
view_shot_video
ChatGPTSHOWS the generated video of an AI Studio shot: extracts REAL FRAMES from the file at the requested seconds (default: 5 frames spread over the whole duration) and returns them as images, along with a public downloadable URL and the duration. Use it to actually see what happens in the video (cuts, motion, how well it matches the prompt). take: 1-based index among the video takes (from get_studio_shots), omitted = active. seconds: up to 8 explicit timestamps in seconds (e.g. [5, 8, 12]).
add_dialogue_alternatives
Claudeadd_outline_beat
Claudeadd_research_note
Claudeadd_scene
Claudechoose_dialogue_alternative
Claudecreate_branch
Claudecreate_project
Claudeget_credits
Claudeget_documents
Claudeget_project
Claudeget_scene
Claudeinsert_scene
Claudelist_characters
Claudelist_projects
Claudelist_research_notes
Clauderename_project
Claudesearch_project
Claudeset_loglines
Claudeset_title_page
Claudeupdate_block
Claudeupdate_document
Claudeupdate_scene
Claude