Starreel

StarReel MCP

Turn a script into a finished, downloadable short-drama episode — from Claude, Cursor, or any MCP client.

[npm version] [npm downloads] [license] [node]

StarReel is a prepaid AI video-production pipeline. This MCP server exposes the whole factory — 120+ tools covering every stage — so an AI agent can take a raw script all the way to a finished .mp4:

script → AI rewrite → cast / scenes / props extraction → storyboards       → character portraits & sheets (anchor the people) + scene plates (anchor the backdrops)       → keyframes → video shots → voiceover (TTS) → final cut (.mp4 link)

It is a thin, open client: all the heavy lifting (character-consistency gates, frame chaining, best-of-N auditing, billing) runs server-side at starreel.ai.

Quick start

[Install in Cursor] [Install in VS Code]

1. Create an API key with produce scope in StarReel → Settings → API Keys (srk_live_..., shown once).
2. Add the server to Claude Code:

claude mcp add starreel -e STARREEL_API_KEY=srk_live_xxx -- npx -y @starreel/mcp

Or install the Claude Code plugin — MCP server + agent skill in one step (set STARREEL_API_KEY in your shell first):

/plugin marketplace add starreel/starreel-mcp/plugin install starreel@starreel

3. Ask your agent: "Take this script and produce a full episode: <your script>" — it will quote each paid stage first and only spend after you confirm.

Tool-catalog language

Tool descriptions and the server instructions are authored in Simplified Chinese. Set STARREEL_MCP_LANG=en for the English edition: the server instructions (the entry-point decision tree) are shipped in English; tool descriptions are being translated in batches and any tool without an English entry yet falls back to Chinese (nothing goes blank — the startup log on stderr reports the current coverage, e.g. lang=en ⚠ English catalog 40/230). The same switch sends Accept-Language so backend-provided labels follow it:

claude mcp add starreel -e STARREEL_API_KEY=srk_live_xxx -e STARREEL_MCP_LANG=en -- npx -y @starreel/mcp

Works with any MCP client

Requires Node ≥ 18 (npx). The standard config below works as-is in Cursor · Windsurf · Cline / Roo Code · Claude Desktop · Trae · Cherry Studio · Chatbox · DeepChat and any client that reads an mcpServers JSON block:

{  "mcpServers": {    "starreel": {      "command": "npx",      "args": ["-y", "@starreel/mcp"],      "env": { "STARREEL_API_KEY": "srk_live_xxx" }    }  }}

Clients with their own format:

[mcp_servers.starreel]command = "npx"args = ["-y", "@starreel/mcp"]env = { "STARREEL_API_KEY" = "srk_live_xxx" }

{  "servers": {    "starreel": {      "command": "npx",      "args": ["-y", "@starreel/mcp"],      "env": { "STARREEL_API_KEY": "srk_live_xxx" }    }  }}

{  "mcpServers": {    "starreel": {      "command": "npx",      "args": ["-y", "@starreel/mcp"],      "env": { "STARREEL_API_KEY": "srk_live_xxx" }    }  }}

What's inside

Project types: drama / ad / mv / brand_film.

Design a voice from a description (no sample needed)

When the narrator or a character has no fitting voice in the library and there is no authorized recording to clone, design one from a single sentence.

The narrator is a character too (char_type voiceover). Describe what you want rather than what you don't — age, gender, timbre, pace, mood — e.g. "a middle-aged man, low and warm voice, slow pace, sentences trail downward". Always make the customer pick one before dubbing: the same description gives a different person every time (measured voice similarity as low as 0.23), while a saved voice is cloned for every line (0.85-0.90). Limits: one design at a time per account, 30 per 24 h; preview sentence 14-28 characters; candidates are kept for 6 hours; never name a real person to imitate their voice.

Billing is agent-safe by design

• Prepaid, never negative. Costs are pre-authorized before any vendor call; insufficient balance returns a clean 402 — nothing half-runs.
• Quote before spend. Big-ticket stages are quote_* → show the user → generate_* with the returned quote_id. For video, quote == actual charge (same function computes both).
• Final cut is free. Composition, transitions, SFX matching and deliverable packaging don't bill.
• API keys are stored hashed and exchanged for 15-minute short-lived tokens; revoke in Settings at any time.

Built for agents: the operating skill

SKILL.md ships inside the package — a platform-agnostic operating manual (full pipeline order + ten operating disciplines + a failure playbook). The server also announces the same guidance at connect time (MCP instructions) and exposes it as a tool, get_capabilities_guide, so an agent that has never seen this file still learns which entry point each kind of material takes. Skill-aware clients load it automatically; on platforms that can't run npx (Coze / Dify / GPTs / custom agents) paste it into the system prompt and drive the same pipeline over REST (/v1/produce/*).

Install it as a standalone agent skill (Claude Code, Codex, Cursor, OpenCode and 70+ agents) via the skills CLI:

npx skills add starreel/starreel-mcp

Environment variables

REST API (OpenAPI)

Prefer plain REST? The full production facade is described in openapi.json (OpenAPI 3.1, 100+ operations — generated from this package's tool surface, so operationIds match MCP tool names 1:1). Browse it rendered at starreel.github.io/starreel-mcp, or generate a typed client for any language:

npx openapi-typescript https://raw.githubusercontent.com/starreel/starreel-mcp/main/openapi.json -o starreel.d.ts

Links

• Website: starreel.ai
• API reference (OpenAPI): starreel.github.io/starreel-mcp
• Full MCP / REST docs: api.shortreelai.com/docs/mcp
• npm: @starreel/mcp
• 中文文档: README.zh-CN.md

License

MIT — this client is open; the production pipeline is a hosted service.