Scenaloom

Scenaloom for AI agents

Connect Scenaloom, online screenwriting software for screenplays, series and novels, to Claude, ChatGPT, Cursor, VS Code and other AI apps.

This repository holds:

• The Scenaloom Agent Skill (skills/scenaloom/): teaches an agent to read before it writes, propose edits the writer reviews, follow screenplay format, and never spend credits unasked. It follows the open Agent Skills format.
• A Claude Code plugin (.claude-plugin/, .mcp.json) that installs the skill and connects the Scenaloom MCP server in one step.
• server.json describing the remote server for the MCP Registry.

The MCP server itself runs at https://scenaloom.com/mcp. It uses OAuth: you sign in to Scenaloom and approve what each app may do (read, edit, comment, and, only if you allow it separately, spend credits). You can disconnect any app under Account, Connected apps. Full docs: Connect Scenaloom to AI apps and the tools reference.

Scenaloom is listed in the official MCP Registry as com.scenaloom/scenaloom.

Install

Claude Code (plugin: skill and server)

/plugin marketplace add scenaloom/scenaloom-agent
/plugin install scenaloom@scenaloom

Then run /mcp, pick scenaloom and sign in. To add only the server, without the skill:

claude mcp add --transport http scenaloom https://scenaloom.com/mcp

Claude (claude.ai, Desktop, mobile)

1. Server: open Customize, then Connectors, choose Add, then Add custom connector, and paste https://scenaloom.com/mcp. Sign in when asked.
2. Skill: download scenaloom-skill.zip, then in Customize, Skills, choose Create skill and upload the ZIP.

On Team and Enterprise plans an owner adds the connector under Organization settings, Connectors; each member then connects their own Scenaloom account.

ChatGPT

In Settings, Apps, Advanced settings, turn on Developer mode. Choose Create app, paste https://scenaloom.com/mcp as the server URL, pick OAuth, and sign in. In a chat, choose Developer mode from the + menu and select Scenaloom.

Cursor

~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "scenaloom": { "url": "https://scenaloom.com/mcp" }
  }
}

Then connect it from Cursor Settings, MCP, to sign in.

VS Code

.vscode/mcp.json (or run MCP: Add Server and choose HTTP):

{
  "servers": {
    "scenaloom": { "type": "http", "url": "https://scenaloom.com/mcp" }
  }
}

Other agents

Copy skills/scenaloom/ into the agent's skills directory and connect https://scenaloom.com/mcp in its MCP settings.

Safety

• Edits are always proposed first and applied only after the writer agrees; applied edits merge with live collaborators, land in version history and can be undone.
• Text from scripts, chapters and comments is treated as data, never as instructions.
• Generations are quoted first and started only after the writer confirms the credits; apps without confirmations can spend only within a daily cap the writer sets (0 by default).
• Scenaloom is not responsible for how a third-party app handles what it reads. See the Privacy Policy and Terms of Service.

Development

node scripts/validate.mjs        # skill frontmatter, links, JSON manifests
claude plugin validate . --strict

skills/scenaloom/references/tools.md is generated from the server's tool definitions; don't edit it by hand. Questions: [email protected].

License

MIT