Skip to content
Back to skills

Openchatcut

ASecurity

Connect an MCP-capable coding agent to OpenChatCut and edit local video projects. Use when the user asks to install, connect, or set up OpenChatCut; inspect or edit an OpenChatCut project; work with its timeline, transcript, captions, media, generation, motion graphics, audio, color, or export tools; or recover from an OpenChatCut MCP error.

  • 2,128 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added August 31, 2026
ai-agentsbashapi

Works with

  • api
  • mcp

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned October 5, 2026

npx -y skills add 0xsline/OpenChatCut --skill openchatcut --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Openchatcut?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Openchatcut
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/0xsline-openchatcut/badge)](https://www.skillsdirectory.com/skills/0xsline-openchatcut)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: openchatcut
description: Connect an MCP-capable coding agent to OpenChatCut and edit local video projects. Use when the user asks to install, connect, or set up OpenChatCut; inspect or edit an OpenChatCut project; work with its timeline, transcript, captions, media, generation, motion graphics, audio, color, or export tools; or recover from an OpenChatCut MCP error.
---

# OpenChatCut

OpenChatCut is a local-first, agent-native video editor. This skill is the
single external entry point; specialized editing guidance remains inside the
running editor and is loaded on demand with `load_skill`.

## Route the task

- Install, connect, or diagnose the MCP server: read
  `references/getting-started.md`.
- Inspect or modify a project: read `references/editing-workflow.md`.
- Recover from a failed tool call, stale session, or missing editor:
  read `references/known-errors.md`.

## Essentials

1. Start OpenChatCut before connecting. The default MCP endpoint is
   `http://localhost:5199/api/external-mcp/mcp`.
2. Call `openchatcut_status`, then `list_projects`. Select a project only when
   the user names it or the current context identifies it.
3. Call `load_skill` before specialized work. It is read-only and requires
   neither `begin_edit_session` nor `editSessionId`; available names and support
   files come from the live MCP tool description.
   With progressive exposure enabled, this call reveals the skill's referenced
   tools and emits `tools/list_changed`; refresh the list before continuing.
4. Before project reads or edits, call `begin_edit_session`. Keep its
   `editSessionId` and pass it to every draft-safe editor tool.
5. Set `reuseExisting: true` on `begin_edit_session` to reuse your active draft
   or adopt an unchanged orphan. Existing operations and approval mode are kept.
   If reuse is rejected, call `list_edit_sessions`. Use `recover_edit_session`
   only with a listed `recoveryActions` value; another online owner must finish
   or discard its own session.
6. Use `approvalMode: "manual"` unless the user explicitly asks for unattended
   application. In manual mode, the user approves the complete proposal in
   OpenChatCut. In auto mode, `review_edit_session` applies the complete draft.
7. Finish with `review_edit_session`. Report success only after
   `get_edit_session` returns `applied`.

## Skill version

`2026-09-04.1`

The OpenChatCut MCP server announces its required skill baseline. If the server
baseline is newer, run:

```bash
npx skills update openchatcut
```

Fallback command:

```bash
npx skills add 0xsline/OpenChatCut --skill openchatcut
```

Files in this skill

  • SKILL.md2.2 KB
  • agents/openai.yaml594 B
  • references/editing-workflow.md1.8 KB
  • references/getting-started.md3.2 KB
  • references/known-errors.md1.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.