CutClip.one

Model Context Protocol · MCP server for video editing

MCP video editor: 65 tools for AI-driven editing

CutClip.one is built on the Model Context Protocol. Creating a scenario, splitting a clip, adding a crossfade, starting a render: each is an MCP tool, and the built-in assistant edits only through them.

Why MCP matters for video editing

The Model Context Protocol is an open standard that lets AI models call tools in a structured, typed way. Instead of a chatbot describing what you should do, an MCP-native editor lets the model do it — through the same validated operations a person would use.

In CutClip.one the user interface, the REST API and the MCP server are thin adapters over one command layer. There is no AI-only backdoor: every mutation is version-checked, attributed to the user or the assistant, and undoable.

The same tools are open to your own AI client. Claude Code, Codex, Cursor or any MCP client that supports Streamable HTTP connects with a personal token, sees only your projects, and spends your credits.

Connect your AI client

Sign in and generate a token in the editor (it fills these in for you), or paste one of these and replace the placeholder.

Run once in a terminal, then ask Claude to work on your Scene Flow projects.
claude mcp add --transport http sceneflow https://www.cutclip.one/api/mcp --header "Authorization: Bearer <YOUR_TOKEN>"

All 65 MCP tools

Read-only tools run immediately; anything that changes the project is proposed for your approval unless auto mode is on.

Projects 8

  • project.create
  • project.list_for_owner
  • project.update
  • project.delete
  • project.get_state
  • project.get_summary
  • project.export
  • project.import

Tracks 4

  • track.create
  • track.update
  • track.delete
  • track.reorder

Scenes 4

  • scene.create
  • scene.update
  • scene.reorder
  • scene.delete

Timeline 17

  • timeline.add_placeholder
  • timeline.move_item
  • timeline.duplicate_item
  • timeline.trim_item
  • timeline.set_speed
  • timeline.set_volume
  • timeline.set_transform
  • timeline.split_item
  • timeline.delete_item
  • timeline.close_gap
  • timeline.close_gaps
  • timeline.batch
  • timeline.add_transition
  • timeline.add_transitions
  • timeline.remove_transition
  • timeline.remove_transitions
  • timeline.update_transition_duration

Assets 4

  • asset.upload
  • asset.replace_placeholder
  • asset.get
  • asset.list

Scenarios 5

  • scenario.create
  • scenario.get
  • scenario.update
  • scenario.delete
  • scenario.apply_to_timeline

Prompts 6

  • prompt.draft
  • prompt.attach
  • prompt.update
  • prompt.get
  • prompt.list_for_scene
  • prompt.list_revisions

Storyline 8

  • storyline.get
  • storyline.set_premise
  • storyline.suggest
  • storyline.set_candidates
  • storyline.pick
  • storyline.update_scene
  • storyline.rewind
  • storyline.apply

Rendering 6

  • render.preview
  • render.final
  • render.status
  • render.get_result
  • render.list_for_project
  • render.delete

History & credits 3

  • project.undo
  • credits.get_balance
  • credits.list_history

What an MCP-native editor gives you

  • Safe autonomy

    Reads happen instantly; edits are proposed as exact tool calls you can approve, reject or auto-apply.

  • Version-checked edits

    Each edit names the project version it expects, so a stale change is rejected instead of overwriting newer work.

  • Batching

    timeline.batch groups many edits into one call, one approval and one undo step.

  • Full-lifecycle coverage

    Idea, scenario, scenes, prompts, timeline, transitions, rendering and undo: the whole workflow is reachable by tools.

  • One command layer

    UI, REST and MCP share the same operations, so behaviour never diverges between them.

  • Auditable history

    Every operation records who made it: a user, the assistant or an external agent.

Frequently asked questions

What is the Model Context Protocol (MCP)?

MCP is an open standard for connecting AI models to tools and data. A model discovers the available tools, calls them with structured arguments and receives structured results.

How many MCP tools does CutClip.one have?

65, across projects, tracks, scenes, timeline editing, assets, scenarios, prompts, rendering, undo and credits.

Can I use these tools from my own MCP client?

Yes. Sign in, generate a personal token in the editor, and add the hosted endpoint to Claude Code, Codex, Cursor or another MCP client. The token only reaches your own projects, actions spend your credits, and you can revoke it at any time.

Are MCP edits safe?

Yes. Edits are validated, version-checked and recorded, the assistant asks for approval by default, and the latest operation can be undone.

Which model uses the tools?

The built-in assistant uses Claude models from Anthropic.

Explore more