Scribis AI Agent Skill & MCP Integration Guide
Learn how to install scribis-transcribe Skill, configure Scribis CLI, and set up Stdio Socket or MCP HTTP Server Port across Windows, macOS DMG, and Mac App Store Sandbox environments.
downloadDownload scribis.install-skill.zip1. Install AI Agent Skill (scribis-transcribe)
Install this Skill into your AI Assistant environments (Claude Code, Cursor, Google Antigravity, Codex CLI, Windsurf) to allow AI to perform local transcription, subtitle correction, and timeline editing automatically.
Step 1: Download ZIP package
Download scribis.install-skill.zip from GitHub Release:
Step 2: Extract to Skills folder
# Windows: %USERPROFILE%\.gemini\config\skills\scribis-transcribe
# macOS: ~/.gemini/config/skills/scribis-transcribe# Project Workspace:
/your-project/.agents/skills/scribis-transcribe{
"entries": [
{ "path": "path/to/scribis-transcribe" }
]
}2. Install & Configure Scribis CLI
Scribis CLI provides a command-line proxy and stdio MCP bridge for AI Hosts.
cd client/cli npm install node bin/scribis.js install-cli
3. Platform & Communication Matrix (Windows / macOS / App Store)
Scribis supports Stdio IPC Socket and MCP HTTP Server channels depending on operating system and distribution channel.
Mode A: Stdio (IPC Socket) Mode
Directly communicates over local high-performance IPC sockets (Windows Named Pipe / Unix Socket) with running Scribis Desktop app. No port configuration needed.
scribis statusMode B: MCP Server Port (HTTP JSON-RPC) Mode
Enable & Usage Steps:
1. Open Scribis Desktop app, go to Settings / MCP Server to check active HTTP Port (e.g. 8890).
2. Pass SCRIBIS_MCP_PORT env var and add --mcp flag in CLI:
SCRIBIS_MCP_PORT=8890 scribis status --mcpWindows (PowerShell):$env:SCRIBIS_MCP_PORT="8890"; scribis status --mcp4. MCP Host (mcp.json) Configuration Examples
Add Scribis service in your Claude Code / Cursor / Windsurf mcp.json configuration:
{
"mcpServers": {
"scribis": {
"command": "node",
"args": ["/path/to/client/bridge.js"]
}
}
}{
"mcpServers": {
"scribis": {
"command": "scribis",
"args": ["transcribe", "--mcp"],
"env": {
"SCRIBIS_MCP_PORT": "8890"
}
}
}
}