JsonCut
Create complete image and video projects visually or with AI, keep every layer editable, and publish one result or every version you need.
https://jsoncut.com/Opens ChatGPT on the web or desktop and asks it to use the WebMCP tools available here.
Connect straight to this server’s public endpoint.
https://mcp.jsoncut.com/mcpWe add this server to your workspace, then open Studio — saved access, one connection to many servers, with a history of what ran.
Last probed Sep 14, 2026 · mcp.jsoncut.com
53tools discovered
Showing 25 of 53 from the live probe.
Jsoncut V2 Get Authoring Guide
Read concise project-structure rules before authoring or restructuring.
Jsoncut V2 Get Primitive Catalog
Discover exact Studio primitives. Controls are editable grouped-layer recipes, not hidden presets. Search icons by at least two characters; returned SVG comes from the Studio Lucide library and must not be redrawn.
Jsoncut V2 List Projects
List durable editable image and video projects.
Jsoncut V2 Create Project
Create one durable HTML-first image or video project. Set source.editabilityTarget to layered, hybrid or atomic when Studio editing granularity matters. Empty variables/customBlocks may be omitted.
Jsoncut V2 Get Project
Read a project including its current source and optimistic version.
Jsoncut V2 Get Project Context
Read one compact project payload with active cut.md guidance and available media, components, templates, brand kits and fonts. Prefer this over several catalogue calls before authoring.
Jsoncut V2 Get Project Cut
Read the latest immutable cut.md version and resolved-reference state.
Jsoncut V2 Validate Project Cut
Validate canonical cut.md Markdown or a typed document without saving it. Provide exactly one representation.
Jsoncut V2 Set Project Cut
Save one validated immutable cut.md version. Use expected_cut_version from get_project_cut; omit it only for the first version.
Jsoncut V2 Get Project Outline
Read a compact paged root timeline and canonical scene ids without returning the large HTML source. Use these ids for later scene calls; optional for long projects.
Jsoncut V2 Get Scene
Read one scene and a compact paged list of descendants. relativeStart matches author_scene; canonical start remains visible for orientation.
Jsoncut V2 Get Project Scope
Read a bounded layer window from a long V2 project. Filters are server-side and full HTML/CSS is not returned.
Jsoncut V2 Get Layer
Read one exact editable V2 layer. Request markup only when a source-level replacement is necessary.
Jsoncut V2 Update Project
Replace a project source after local validation; prefer bounded edits for small changes.
Jsoncut V2 Edit Project
Apply one bounded group, ungroup, reparent, timing or reorder operation.
Jsoncut V2 Author Project
Apply bounded editable layers, optional scenes, SVG, custom blocks or keyframes. Prefer add_text/add_shape/add_media for designed compositions; use SVG as an intentional hybrid/atomic layer when appropriate. Text inside compact controls must share their bounds and be centered.
Jsoncut V2 Author Scene
Edit one long-video scene by its canonical outline id with <=24 typed actions. Times are scene-relative. Send only one mutation call per response, then use its returned version.
Jsoncut V2 Patch Project Source
Apply one complete, bounded HTML/CSS fragment. Prefer semantic author/edit tools for independently editable objects. Atomic SVG/HTML remains allowed; validate_project reports reduced editability without blocking it unless source.editabilityTarget is layered.
Jsoncut V2 Set Keyframes
Set editable Studio keyframes on one layer. Times are clip-local seconds. Use scene_id only for a layer inside an optional scene; position values are {x,y}, never separate x/y tracks. Keep primary content visible at frame zero and span a meaningful share of the clip instead of ending all motion in its first second.
Jsoncut V2 Validate Project
Preflight without rendering. Defaults infer image/PNG or video/MP4; normally only project_id is needed.
Jsoncut V2 Check Project
Optionally create a persisted, version-bound quality review. This works without cut.md and never gates ordinary rendering. Evidence accepts completed exact-frame renders, including renderer-native crops. Reference comparisons require a frame render, a cut.md-linked image media ID and an optional normative similarity threshold.
Jsoncut V2 Get Review
Get persisted findings, evidence, staleness and approval state for one review or the latest review.
Jsoncut V2 Set Review Status
Record an optional Agent review decision for the exact project and cut versions. Human approval remains a Studio/API-only decision and ordinary rendering needs neither.
Jsoncut V2 Review Scene
Queue a contact-sheet review sampled only across one scene; inspect it later with get_render(include_artifact=true).
Jsoncut V2 Review Playback
Queue exactly one silent draft video for complete motion, transition, timing and text-effect playback review. After completion inspect it with get_render(include_artifact=true), watch the entire artifact, then persist the result with check_project(full_playback_confirmed=true, render_id=...).
Get your MCP into directories
A working endpoint is step one. Directory coverage is the coordinated launch across ChatGPT, Claude, Cursor, the MCP Registry, and community indexes.
Directory coverage for brandsCreate complete image and video projects visually or with AI, keep every layer editable, and publish one result or every version you need.
Use the MCP endpoint listed on this page in your MCP client configuration. One-click install pills support Claude, Cursor, VS Code, and other hosts. Copy the remote MCP URL if your client needs a manual entry.
Operate JsonCut? Verify ownership to take over this directory entry.
This server appears in the MCPBundles directory. Verify you operate it to take over the listing — name, description, logo, contact email, and skill content. We email a 6-digit code to a maintainer address your server publishes in /.well-known/security.txt or /.well-known/mcpbundles.json. Free, takes about a minute.
MCPBundles probed 53 tools on the live server. The tool list on this page reflects what was discovered at the last refresh — connect your client to see the full set available to your session.
No provider sign-in was required during MCPBundles' probe. Your client may still need MCPBundles credentials depending on how you connect.
MCPBundles is an independent platform built on the open Model Context Protocol standard. Not affiliated with Anthropic PBC or Claude.