Olo Studio MCP v2
@olo-studio/studio-mcp · v1.3.1 SceneSpec v2 · Cinema Engine

Your agent now directs 3D cinema
in Claude Code.

Give Claude Code, Antigravity, Codex, or Cursor a complete studio with 10 cinema cameras, 15 optical Post-FX, 16 3D gestures, and local 9:16 / 16:9 / 1:1 / 4:5 renders at $0 cloud GPU.

INSTANT INSTALL · CLAUDE CODE
Skill + MCP Server stdio
claude mcp add olo-studio -- npx -y @olo-studio/studio-mcp && npx skills add faustodariopenamondragon-web/OlomniAI --skill olo-studio
10 CINEMA CAMERAS 15 POST-FX + 5 LUTS 10 TRANSITIONS 14 HUD CARDS 8 MCP TOOLS $0 CLOUD GPU
DIRECT ANSWER · AEO / GEO

What is @olo-studio/studio-mcp and SceneSpec v2?

@olo-studio/studio-mcp (v1.3.1) is a local-first stdio MCP server that connects AI coding agents (Claude Code, Antigravity, Codex, and Cursor) to Olo Studio. Using the deterministic SceneSpec v2 (olomni.studio.scene.v2) contract, agents compile and render 60 fps 3D videos (1080×1920 MP4, WebM Alpha, Chroma MP4) locally in Chromium at $0 cloud GPU cost.

01 / LIVE CINEMA BENCH

From prompt to cinematography.

Pick an engine preset: your agent queries the olomni.studio.scene.v2 catalog, directs camera, lighting, HUD cards, and visually inspects frames before delivering your .mp4.

CAM: DOLLY_ZOOM_VERTIGO · POST-FX: DOF_BOKEH + HALATION

3s hook with anamorphic camera, HUD cards & kinetic captions

@OloStudio Create an 18s 9:16 Reel announcing our v2.4 launch with whip_zoom_in → dolly_zoom_vertigo camera, dof_bokeh, a metric card, and previewFrames: true.
olomni.studio.scene.v2 · Live agent payload
{
  "schema": "olomni.studio.scene.v2",
  "aspect": "9:16",
  "typography": { "style": "kinetic_slam", "highlightStyle": "underline" },
  "postFx": ["dof_bokeh", "halation", "motion_blur", "vignette"],
  "beats": [
    {
      "speaker": "Olo",
      "text": "Stop paying for cloud GPU renders.",
      "camera": "dolly_zoom_vertigo",
      "gesture": "double_point",
      "transition": "whip_pan",
      "hudCard": { "type": "metric", "data": { "label": "GPU Cost", "value": "$0.00" } }
    }
  ]
}
Try in /studio
02 / ENGINE v2 CAPABILITIES

High-level cinema tokens as black-box blocks.

Your AI agent selects declarative tokens via olo_describe_capabilities. The hardened local engine executes continuous 3D spring physics, anamorphic optics, and DSP audio without exposing internal source code.

CINEMA CAMERAS & RIGS 10

Deterministic optics without jump cuts

From Hitchcock-style vertigo zooms and FPV swoops to rising crane moves and organic handheld framing with continuous beat interpolation.

dolly_zoom_vertigo fpv_swoop crane_rise whip_zoom_in rack_focus_push orbit_reveal hero_low_angle dutch_tension handheld_ugc static
OPTICAL POST-FX & COLOR 15

Film-lab grade optical texture

Shallow-aperture bokeh, warm Kodak halation, velocity-driven radial motion blur, anamorphic flares, and 5 cinematic color LUTs.

dof_bokeh halation motion_blur film_gate anamorphic chromatic_aberration color_lut light_leak_warm cinematic_bars film_grain golden_dust vhs_retro
TRANSITIONS 10

Rhythm-matched cuts

Includes slide_push, flash cuts, chromatic glitches, editorial curtains, and warm light leaks.

slide_push whip_pan iris_wipe zoom_punch glitch_cut light_leak curtain_reveal
3D GESTURES + IK 16

Spring-damped continuous arms

Second-order damped springs deliver smooth transitions across every beat and emotional state.

double_point inspect_mockup open_present mind_blown countdown_hold mic_drop celebrate
HUD, MOCKUPS & AUDIO 14+

Ad-ready retention overlays

14 HUD cards, 4 auto-scrolling device mockups, 5 typography presets, 16 foley cues, and 9 music genres.

price_slash social_proof_pill iphone_16 macbook_pro kinetic_slam
03 / API & MCP ARCHITECTURE

7 MCP tools. One frictionless autonomous loop.

Your agent discovers the catalog, compiles a deterministic storyboard, renders in local Chromium with previewFrames: true to visually verify frames, and delivers the final .mp4 + .srt.

01 · DISCOVER

Call olo_describe_capabilities

Fetch the live 10 camera presets, 15 Post-FX, 16 gestures, 14 HUD cards, and SceneSpec v2 starter recipes.

02 · COMPILE

Build olomni.studio.scene.v2

Use olo_compile_scene or pass specs directly to olo_render_video with brand tint, device mockups, neural voice, and synced foley.

03 · VERIFY & EXPORT

Render + previewFrames: true

Saves JPEG keyframes alongside the video so your agent validates framing and captions before delivering the final MP4 or .ZIP.

DISCOVERY olo_describe_capabilities

Live manifest of cameras, Post-FX, transitions, gestures, HUDs, mockups, and SceneSpec v2 recipes.

SCRIPTING olo_generate_script

Turns URLs, changelogs, or briefs into Olo: / Yo: beat scripts with hooks and offline fallback.

COMPILER olo_compile_scene

Validates and compiles deterministic olomni.studio.scene.v2 / v1 JSON without launching a browser.

3D RENDER olo_render_video

Exports locally in Chromium as MP4, WebM Alpha, or Chroma MP4 (9:16, 16:9, 1:1, 4:5) + .srt.

A/B SCALE olo_render_batch

Batch-renders up to 50 variations from CSV or JSON into a single batch.zip with manifest and subtitles.

VOICE (BETA) olo_voice_clone_list

Lists and manages custom AI voices tied to your account (Beta · Coming Soon).

AUTH & TIER olo_get_entitlements

Inspects your active tier, export quotas, and feature flags via your olo_pat_... token.

Personal Access Token Auth (olo_pat_...) Free without a key for local sandbox renders. To unlock zero watermark, 3D Outro, .GLB auto-rig, and Pro/Agency batches, create your PAT in /studio.
Create olo_pat_ token in /studio
Universal JSON config for Claude Desktop, Windsurf, Antigravity, or any MCP client
{
  "mcpServers": {
    "olo-studio": {
      "command": "npx",
      "args": ["-y", "@olo-studio/studio-mcp"],
      "env": {
        "OLO_STUDIO_API_KEY": "olo_pat_your_token"
      }
    }
  }
}
04 / COPY-PASTE AGENT PROMPTS

Ask in plain English. Get the render.

Click any workflow card to inspect the exact prompt your coding agent runs once connected to @olo-studio/studio-mcp.

Claude Code: “Inspect olo_describe_capabilities and render a 20s 9:16 Reel with Olo announcing our v2.4 launch using dolly_zoom_vertigo, dof_bokeh, and kinetic_slam captions with previewFrames: true.”
Open in /studio
Open Beta · Early Access

Real plans. Zero vaporware.

Start free today with the local MCP server. Activate Pro or Agency via priority access in /studio.

Sandbox

Free

$0 / mo
No credit card · Ready in 10 seconds
  • 3 exports/day · 9:16 up to 30 s · #MadeWithOlo watermark
  • Unlimited Edge Neural HD voices ($0)
  • Local MCP SceneSpec v2 compile & render
Start free
Open Beta · Teams

Brand & Agency

$149 / mo
Annual: $119/mo ($1,428/yr) · Save 20%
  • 500 exports/mo · Custom .GLB/.GLTF auto-rig · up to 10 min (600 s)
  • Bulk CSV / MCP 50 ads .ZIP · 5 seats · Paid Ads license
  • Edge Neural HD voices included · More AI voices coming soon (Beta)
Request access in /studio

Open Beta: the Free sandbox and local MCP engine work today with no credit card; Pro and Agency Stripe billing is enabling soon via priority access in /studio. Full machine-readable sheet at pricing.md.

06 / TECHNICAL COMPARISON

How @olo-studio/studio-mcp compares.

Factual metrics comparing Olo Studio against Remotion, HeyGen API, and Runway for automated agent-driven video pipelines.

Metric @olo-studio/studio-mcp (v1.3.1) Remotion HeyGen API Runway
Runtime & Cloud GPU Local Chromium WebGL2 + WebCodecs ($0 cloud GPU) React + Chromium / FFmpeg (local or paid Lambda) Cloud render (per-minute credits) Cloud diffusion (per-second credits)
AI Agent Protocol Native stdio MCP (7 tools) + Universal Skill Custom wrappers or React code REST API REST / Web API
Cameras & Post-FX 10 cinema cameras, 15 Post-FX + 5 LUTs, 10 transitions Manual CSS / Three.js code Fixed talking-head framing Prompt-dependent motion
Deterministic Output 1080×1920 (9:16, 16:9, 1:1, 4:5) 60 fps + WebM Alpha + .srt + previewFrames Deterministic 2D / React Cloud MP4 (no local WebGL alpha) Non-deterministic diffusion clips
Bulk CSV & .GLB Mascots Up to 50 ads in .ZIP + auto-rig .GLB/.GLTF Self-built Lambda pipeline Streaming Avatar SDK N/A
Entry Pricing Free $0 · Creator & Builder Pro $19/mo · Brand & Agency $149/mo Free OSS (paid Lambda/team license) Paid credit tiers Paid credit tiers
07 / FREQUENTLY ASKED QUESTIONS

Technical answers, plainly stated.

What is @olo-studio/studio-mcp and SceneSpec v2?

@olo-studio/studio-mcp (v1.3.1) is Olo Studio's local-first stdio MCP server. It exposes 7 tools so Claude Code, Antigravity, Codex, and Cursor can compile and render SceneSpec v2 (olomni.studio.scene.v2) storyboards with 10 cinema cameras, 15 Post-FX, 16 3D gestures, and 14 HUD cards locally in Chromium.

Why does rendering cost $0 in the cloud, and what is the maximum resolution?

The entire WebGL2 engine and WebCodecs encoder run on your own machine inside local headless Chromium, with zero remote GPU servers. Maximum export resolution is 1080×1920 in 9:16 (plus 1920×1080 in 16:9, 1080×1080 in 1:1, and 1080×1350 in 4:5) at up to 60 frames per second.

How does olo_pat_ authentication work, and what does the Free tier include?

The Free sandbox ($0/mo) and SceneSpec v2 compiler work with no account or key (9:16 clips up to 30 s with a discrete watermark and unlimited Edge TTS). To unlock Creator & Builder Pro ($19/mo) or Brand & Agency ($149/mo), issue a Personal Access Token (olo_pat_..., SHA-256 hashed, max 10 active keys) in /studio and set OLO_STUDIO_API_KEY.

Is SceneSpec v2 backward-compatible with v1, and how does previewFrames prevent visual bugs?

Yes. @olo-studio/studio-mcp v1.3.1 stays compatible with olomni.studio.scene.v1 and introduces previewFrames: true on olo_render_video to extract up to 6 JPEG keyframes so your AI agent visually inspects framing, HUD cards, and captions before returning the final video.

What is the difference between Creator & Builder Pro and Brand & Agency?

Creator & Builder Pro ($19/mo or $15/mo annual, $180/yr) unlocks 60 exports/mo, 4 ratios, zero watermark, custom 3D Outros, WebM Alpha and Chroma MP4, clips up to 3 min (180 s), Local-First MCP, and unlimited Edge Neural HD voices (More AI voices & cloning: Beta · Coming Soon). Brand & Agency ($149/mo or $119/mo annual, $1,428/yr) expands to 500 exports/mo, 10 min (600 s), custom .GLB/.GLTF auto-rigging, 50-ad Bulk CSV / MCP .ZIP batches, Paid Ads commercial license, 5 team seats, and unlimited Edge Neural HD voices (More AI voices & cloning: Beta · Coming Soon).