animate_from_card
ChatGPTUsed by the Fenestra render card to start an animation from an image it uploaded.
animate_image
ChatGPTFenestra architectural video. Turns a render or photo of a building, room, facade or landscape into a short cinematic clip (5 seconds) with a camera move: dolly in or out, orbit, pan, zoom, drone rise, or static. Use it for "animate this", "make a video, walkthrough or flythrough", or "add camera movement". After a Fenestra render with no new image attached, pass that render's image_url. To transition between two images (for example a day render and its evening version), pass the first as the image and the second as end_image_url; camera_move "static" suits a time-of-day transition. The prompt describes motion and atmosphere in the scene (people walking, trees in the wind, clouds drifting, lights coming on), never a redesign. Uses the user's Fenestra credits and saves the clip to their Fenestra project. If the user attached an image to their latest message, always pass that image as image, even when earlier renders exist. Only when the latest message has no image and refers to an earlier result ("now make it dusk"), pass the Fenestra URL of that result as image_url. Only if there is no image in the chat at all, call this with just the prompt: Fenestra shows an upload card.
create_upload
ChatGPTOnly for coding agents (e.g. Claude Code, Cursor) uploading a file from the user's own machine. Not for chat apps: there, call render_image with just the prompt and Fenestra shows the user an upload card. Returns a one-time URL; upload with the returned curl command (HTTP PUT, exact Content-Type), then call render_image with image_url.
enhance_from_card
ChatGPTUsed by the Fenestra render card to enhance an image it uploaded.
enhance_image
ChatGPTFenestra image enhancement: upscale and sharpen an architectural or interior image to 2K or 4K. Use it for "enhance", "upscale", "sharpen", "make it higher resolution" or "print quality". The default keeps the design exactly as it is; use style "add_detail" only when the user wants more realism or texture. Works on Fenestra renders or any photo. Uses the user's Fenestra credits and saves the result to their Fenestra project. If the user attached an image to their latest message, always pass that image as image, even when earlier renders exist. Only when the latest message has no image and refers to an earlier result ("now make it dusk"), pass the Fenestra URL of that result as image_url. Only if there is no image in the chat at all, call this with no image: Fenestra shows an upload card.
get_credits
ChatGPTHow many Fenestra credits the user has left for renders.
get_render
ChatGPTCollect a Fenestra render that was still in progress, using the predictionId render_image returned. Waits up to 25 seconds for it to finish.
get_video
ChatGPTCollect a Fenestra animation that is still in progress, using the predictionId animate_image returned. Waits up to 25 seconds for it to finish.
list_projects
ChatGPTThe user's Fenestra projects, most recently updated first, and which one new renders go to. Use it when the user wants renders in a particular project; then call set_project with the id.
list_workspaces
ChatGPTThe user's Fenestra workspaces (personal and team) and which one is active. Renders spend the active workspace's credits and go to its projects. Use it when the user mentions a team or workspace; then call set_workspace with the id.
render_from_card
ChatGPTUsed by the Fenestra render card to start a render from an image it uploaded.
render_image
ChatGPTFenestra architectural and interior rendering. Use it for any edit or render of a building, room, facade, landscape or space: make it evening, dusk, night or golden hour; change the sky or weather; swap materials, cladding or finishes; add or remove furniture, people, trees or landscaping; restyle an interior; turn a sketch, SketchUp, Rhino, Revit or Archicad screenshot, massing model or site photo into a photoreal render. It keeps the architecture, geometry and camera intact and saves the result to the user's Fenestra project. The prompt describes the finished still: light, materials, composition and atmosphere. Never put camera movement (orbit, dolly, pan) in it. Uses the user's Fenestra credits and saves the render to their Fenestra account. If the user attached an image to their latest message, always pass that image as image, even when earlier renders exist. Only when the latest message has no image and refers to an earlier result ("now make it dusk"), pass the Fenestra URL of that result as image_url. If there is no image you can pass, call this with just the prompt: Fenestra shows an upload card that uploads the image and starts the render. Do not try to upload it yourself first.
roast_design
ChatGPTFenestra design roast. Call this when the user asks to roast, rate or judge an architectural, interior or landscape design. It returns Fenestra's roast format and voice: write the roast, then call render_image with the same image and the roast, so the card shows it above the fixed design.
set_project
ChatGPTSets the Fenestra project new renders, enhancements and videos go to, in this and later chats. Pass a project id from list_projects, or "default" for the default project.
set_workspace
ChatGPTMakes a workspace from list_workspaces the active one, in Fenestra as well as here. New renders then spend its credits and go to its default project.
upload_image_data
ChatGPTUsed by the Fenestra render card to upload the image the user picked.
whoami
ChatGPTThe Fenestra account this connection is signed in as.