studio-mcp

The block-method film pipeline as agent-callable MCP tools.

An MCP server that lets an LLM agent drive a full brief → shots → render → QC → cut workflow by calling tools — instead of clicking through generation UIs by hand. Plan-then-generate, never reverse: every shot is planned (type, move, duration) before any pixel is made, and every still is QC'd against a locked look so the piece stays on-model.

Why

Single-tool MCP servers (one model, one call) are common. This is the layer above: it orchestrates across your generation stack with a block-method planner and a style-drift QC gate — the part nobody ships.

Related MCP server: Media-infrastructure

Tools (v1)

v1.1 (render phase, via Higgsfield CLI): train_character (soul-id self-clone) · gen_still (Soul) · animate (Seedance/Kling).

Provider-agnostic

The LLM layer talks to any OpenAI-compatible endpoint — Anthropic, OpenRouter, OpenAI, or a local server — chosen entirely through env config. No provider is hard-coded.

cp .env.example .env   # set STUDIO_LLM_BASE_URL / _MODEL / _API_KEY

Install

python3 -m venv .venv && source .venv/bin/activate
pip install -e .

Run

studio-mcp        # stdio MCP server

Register with an MCP client (e.g. Claude Code / Claude Desktop):

{
  "mcpServers": {
    "studio": {
      "command": "/path/to/studio-mcp/.venv/bin/studio-mcp",
      "env": { "STUDIO_LLM_API_KEY": "sk-..." }
    }
  }
}

Example flow

plan_shots("MARÉA — wordless coastal slow-burn, 90s", project="marea")
lock_campaign("marea", aspect="2.39:1", stock="Kodak 500T, soft handheld",
              hex_palette=["#1b2a3a","#c8a15a"], elements=["the woman in grey"],
              audio="diegetic SFX only, no music")
# render a still (v1.1) → then:
qc_still("marea", image="assets/shot1.png", shot_id=1)   # pass/fail + fix
assemble("marea")                                         # cut manifest

State lives under STUDIO_ROOT (default ~/studio-projects/<project>/) as plain JSON — human-inspectable, and a clean contract a future console can read.

License

MIT