Picorn MCP

What Picorn MCP MCP server does

Picorn MCP MCP server lets an MCP client create and manage visual work in Picorn. Its scope covers layered images, persistent editor projects, multi-slide files, and timeline-based videos. Results can be rendered to public image URLs, PDFs, DOCX files, or finished MP4 and WebM videos, depending on the workflow.

The server supports both agent-authored and model-assisted asset creation. Agents can provide complete SVG layouts, including text, shapes, and declared image placeholders, or request raster images and vector assets from Picorn's hosted generation pipeline. Projects can then be opened in Picorn for manual changes.

How it works

Start by checking capabilities when the workflow depends on available canvas formats, fonts, layouts, image models, GPU status, or fair-use limits. For marketing projects, create a project and retain its workspace project ID and key together when adding files. Project inspection returns editor files, imported assets, brand information, thumbnails, statuses, and canonical editor URLs.

For native editing, create a project or slideshow from completed SVG and editable layers. The server can resolve declarative SVG placeholders, store generated assets, and convert the result into editable native layers. Incremental workflows can import external image bytes, add or update layers, arrange them, and render selected or all slides.

Video workflows use a persistent 30fps timeline. Agents can create a video project, import supported media, add clips, configure timing and trims, and use keyframes or timed layers before rendering to MP4/H.264 or WebM/VP9.

Tools and capabilities

The Picorn MCP MCP server exposes tools for:

• Reading live capabilities and inspecting project contents.
• Creating and updating marketing projects, editor projects, slideshows, and video projects.
• Generating raw images with self-hosted Z-Image Turbo and editing images with Flux 2 Klein.
• Creating vector assets through raster generation, background removal, and multi-colour SVG conversion.
• Rendering SVG layouts with selectable image backends and declarative placeholders.
• Importing image and video/audio media into Picorn storage.
• Adding, updating, deleting, duplicating, grouping, ungrouping, aligning, distributing, and reordering editor layers.
• Adding, updating, and deleting slides, with editor projects supporting up to 20 slides.
• Rendering image projects to images, PDF, or DOCX, and rendering timelines to MP4 or WebM.

Layer operations support text, image, shape, group, QR, video, and audio types. Persisted properties include transforms, styles, filters, masks, effects, blend modes, visibility, timing, audio settings, and keyframes where applicable.

Limitations and notes

The design-image tool is Codex-only and expects the calling Codex agent to provide the complete SVG; it does not call an internal language or design model. The slideshow path likewise uses agent-authored SVG or native layers and reports the calling agent as the model provider.

Video and audio imports accept MP4, WebM, MOV, MP3, WAV, AAC, M4A, and OGG files up to 15 MB. Video clips must reference an approved public media host. Slideshow creation is limited to 20 slides, 50,000,000 total canvas pixels, and 1,000 total layers. Fair-use slideshow cost is one unit per four slides, rounded up.

The service returns public creation or render URLs and editor links, so clients should present the canonical editor URL when a workflow produces one. No API key is required according to the supplied description.