Openclaw Ai Video Editor MCP Server

Agentic AI Video Editor

An autonomous agentic video editor. Send a sentence. Get a finished video. No timelines, no keyframes, no plugin chains. Get hundreds of reels and edits rendered within seconds using massive parallel GPU rendering infrastructure.

Beta. The agent can make mistakes. Preview every output before publishing or sharing. For high-stakes or irreversible edits, pass requirePlanApproval: true so the agent stops after planning and waits for your approval before anything runs.

---

Contents

1. Quickstart — running in 60 seconds
2. What it does — say it, get it
3. Capability surface — the full toolset
4. Renderer & technical architecture — Vulkan GPU compositor, shader pipeline, MG library
5. Goals & briefs — from one line to a full creative brief
6. API reference — endpoints, request/response, streaming, async
7. Who it's for · vs. legacy editors · Safety · Media
8. Links & channels

---

Quickstart

The portable interface is the levea-mcp-server MCP server — one server, every MCP client (Claude Desktop, Claude Code, Cursor, Cline, OpenClaw, Hermes), one tool surface, one backend contract, so nothing drifts per platform.

1. Get an API key — sign up at studio.livecore.ai and generate an OpenClaw API key.

2. Add the MCP server — the same npx line works for every MCP client:

{
  "mcpServers": {
    "levea": {
      "command": "npx",
      "args": ["-y", "levea-mcp-server"],
      "env": {
        "LEVEA_API_URL": "https://api.livecore.ai",
        "LEVEA_API_KEY": "your-key-from-studio.livecore.ai"
      }
    }
  }
}

3. Use it — say to your AI: "Use the levea tool to generate 5 viral clips from this video, add captions, reframe vertical." The agent picks the tool, fills the prompt; the editor plans → executes → exports.

Per-client setup

This listing on ClawHub: openclaw plugins install clawhub:openclaw-ai-video-editor

Every client calls the same backend — the MCP server, this ClawHub listing, and every other surface share one tool contract.

Two hosts — don't confuse them

Env vars are LEVEA_API_URL / LEVEA_API_KEY.

Tools the AI sees

One editing entry point. Everything else is typed state-management and polling — so integrators can build full editing experiences through one API key instead of mixing JWT and MCP.

No edit shims (add_captions, generate_viral_clips, …) and no editor_execute escape hatch. The backend's planner is the specialist — fragmenting the editing surface encourages calling LLMs to second-guess the planner and lose multi-step intent. State-management tools (brand / project / asset / caption-template CRUD) aren't editing, so they don't fragment intent — they just give integrators typed access.

→ For agent-integration details (schemas, async semantics, error handling, plan approval), see AGENTS.md.

---

What it does

Say it in plain language; the agent plans and finishes the edit.

---

Capability surface

Timeline & Editing Core

• Multi-track timeline with layer groups, locking, visibility toggles, snap guides, zoom, scrubbing
• Trim, split, crop, speed ramps, freeze frame, reverse clip
• Keyframe animation — position, scale, rotation, opacity, blur, volume — with easing curves
• 17 Photoshop-accurate blend modes · masking (luma, alpha, chroma, depth, LUT)
• Canvas configuration — dimensions, aspect, FPS
• Batch updates · undo/redo · scene save/load/duplicate · final reconcile

Captions & Text

• Generate captions from transcript; restyle or translate existing ones
• 25+ built-in caption templates (Hormozi, MrBeast, neon-pop, word-pill, karaoke, explainer-burst, and more)
• Import SRT / VTT · caption repair · filler-word removal · safe-zone repair
• Title cards, lower-thirds, end cards / CTAs · chapter markers · text on a path
• MSDF/SDF rendering — sharp at any size, 10 visual modes (plain, outline, shadow, glow, neon, striped, and more)
• Per-character animation: typewriter, slide, fade, scale, bounce, flip, rotate, karaoke word-highlight

Audio & Voice

• AI voiceover (Google Cloud TTS) — 40 languages · multi-language dubbing/translation
• Voice cloning — custom voices from reference + consent recordings
• AI music generation (Lyria, prompt-driven) · sound effects generation
• Audio mastering to platform targets (YouTube / Spotify / TikTok / Podcast / Cinema, EBU R128 LUFS)
• Parametric EQ (8 bands) · mixing & dynamics · crossfades (J/L-cut) · auto-ducking (4 modes)
• Silence removal · beat-sync editing · stem separation · mute/solo

Color, VFX & Visual Treatment

• LUT color grading + 3-way color wheels (lift/gamma/gain) + 12 cinematic preset looks
• 21 procedural VFX: smoke, fire, explosion, portal, lightning, lava, snow, bokeh, glitch, decay/corrosion, scanlines/CRT/VHS, film grain, glassmorphism, halftone, liquid/gooey, holographic, kaleidoscope, dithering, tunnel/wormhole, aurora/mesh gradient, light sweep
• AI neural background removal — no green screen required
• Chroma key / green screen · branding overlays · face/motion tracking with dynamic zoom
• Video stabilization · noise reduction · depth-map driven effects
• Waveform / vectorscope scopes

AI Content Generation

• Text-to-image / text-to-video (Imagen, Veo) · image-to-video · extend-video
• B-roll generation — contextual, transcript-grounded
• Motion graphics — 128+ animated types (see Motion Graphics Library)
• Thumbnail generation · generate-from-brief · slideshow (Ken Burns)
• Frame analysis — deep frame-by-frame visual analysis

Charts & Data Visualization

• Bar grow, line reveal, pie segment, icon array waffle, dashboard (multi-panel), bar race, sankey flow, treemap
• Auto-detect chartable data in narration and build the chart
• Brand-aware styling · canvas-responsive

Viral / Social & Multi-Platform

• Viral clip selection — find best moments; export combined reel or separate clips
• Vertical reframe to 9:16 with face-aware tracking
• Multi-platform export — TikTok, Reels, Shorts, YouTube, LinkedIn, Instagram in one pass
• Split-screen / PIP · multi-cam sync · cross-asset face/speaker search

Privacy & Compliance

• Face blur · privacy redaction · profanity cleanup · object hide
• Content-safety gating — blocks deepfakes, fake testimonials, unauthorized likeness

Export & Delivery

• H.264, H.265, WebM · 4K UHD, 1440p, 1080p, 720p, 480p
• TikTok Pro, TikTok/Reels, Square, IG-portrait presets
• SRT / VTT subtitle export · pre-export delivery check
• Up to 3 hours per asset

---

Renderer & Technical Architecture

Everything below is implemented in production — not planned. The renderer is the moat: no competitor in this space runs a real GPU compositor server-side.

Architecture

User prompt / API call
  → LLM emits typed IR (intent + params)
  → Deterministic DAG compiler (60+ task kinds, 118+ canonical actions)
  → Gated executor → PowerPacks
  → Verifier → bounded repair loop (max 2 attempts)
  → Vulkan GPU compositor → NVENC → GCS export

Preview runs on WebGPU in the browser. Export runs on Vulkan server-side. Both surfaces share the same shader source — compiled to SPIR-V for Vulkan and consumed natively by WebGPU — so preview and export output are pixel-identical.

Hardware pipeline: NVDEC video decode → CUDA → Vulkan zero-copy texture pool (ML inference results feed compositor directly) → NVENC H.264/H.265 encode.

Post-Processing Shader Pipeline

18-pass compositor stack, each pass independent and keyframe-controllable:

SSGI is not a standard video editor feature. DaVinci Resolve ships it as a premium Fusion node; After Effects requires third-party plugins. It's a built-in shader pass here.

VFX Effects (21)

All keyframe-animated with intensity, speed, size, spread, seed controls:

smoke/dust/steam/mist/fog · fire · explosion/glimmer · portal · lightning · lava · snow · bokeh · glitch · decay/corrosion · scanlines/crt/vhs · film_grain · glassmorphism/frosted_glass · halftone · gooey/liquid · holographic/iridescent · kaleidoscope · dithering/bayer · tunnel/wormhole · aurora/mesh_gradient · light_sweep/shine

GPU Compute Particle Physics

Physics simulated on GPU via compute shaders — real particle dynamics, not sprite animations:

Rain · Snow · Dust / Embers · Portal vortex — each particle system runs velocity, drag, gravity, and lifetime simulation on the GPU per frame.

Transitions (33)

All directional (8 directions). Optional neon edge-glow composited on the seam.

Basic: fade · slide · wipe · iris · zoom · blur fade · cross dissolve · whip pan · zoom blur

Cinematic: dip to black · dip to white · luma fade · glitch · spin blur · spin · RGB split · VHS static · film dissolve · glow dissolve · directional blur

Geometric / Pattern: radial wipe · barn door · checkerboard · gradient wipe · venetian blinds · star wipe · bowtie wipe · pixelate dissolve · ripple warp · lens distort · polka dots · perlin dissolve · kaleidoscope

Blend Modes (17)

Normal · Multiply · Screen · Overlay · Darken · Lighten · Color Dodge · Color Burn · Hard Light · Soft Light · Difference · Exclusion · Hue · Saturation · Color · Luminosity · Additive

3D Transforms

Per-layer 3D transforms: perspective camera · rotate X/Y/Z · translate Z (depth) · parallax 2.5D camera push · 3D card flip · extruded 3D text

Text Rendering (MSDF/SDF)

Pixel-perfect at any size. 10 SDF visual modes: plain · outline · shadow · outline+shadow · outer glow · inner glow · dual glow/neon · striped · landscape gradient · overload (multi-color)

Per-character animation: typewriter · slide up/down/left/right · fade in · scale in · rotate in · bounce in · flip in · karaoke word-highlight (transcript-synced)

Animation Runtimes

Skottie / Lottie — rendered via Skia on the Vulkan GPU backend. Full spec: trim paths, repeaters, blend modes, expressions. GPU-accelerated, zero-copy path from decoder to compositor. 158 bundled accent-grade assets. LLM-generated concept Lotties via procedural builder. SVG → Lottie transpiler (nested groups, transforms, opacity).

Rive — state machine + timeline animations, GPU-accelerated on Vulkan.

AI Perception (GPU)

All inference runs on GPU. Capabilities:

---

Motion Graphics Library

128+ motion graphic types — all LLM-selectable, all animated, all rendered by the Vulkan compositor.

Charts & Data Visualization (8)

chart.bar_grow · chart.line_reveal · chart.pie_segment · chart.icon_array_waffle · chart.dashboard · chart.bar_race · chart.sankey_flow · chart.treemap

Diagrams & Process (5)

diagram.explainer · diagram.schematic · diagram.equation · process.flowchart · process.decision_tree

Lists & Comparisons (5)

list.numbered_steps · list.bullet_stagger · compare.vs_card · compare.split_screen · compare.checklist

Stats & Data Callouts (8)

stat.counter · stat.big_number · stat.percentage_ring · stat.comparison_bar · stat.trend_arrow · stat.growth_chart · stat.leaderboard · stat.multi_metric

Captions & Kinetic Subtitles (4)

caption.word_pop · caption.keyword_highlight · caption.supersize · caption.emoji_burst

Kinetic Typography & Titles (15)

text.paragraph · text.word_swap · text.echo_stack · text.on_path · text.typewriter_prompt · text.glitch_reveal · text.stretch · text.mirror · text.cascade · text.neon_glow · text.word_reveal · title.section_card · title.mega_type · title.blend_overlay · title.extrude_3d

Product / SaaS / Pitch Deck (8)

mockup.browser · mockup.phone · product.callout · product.hero_reveal · pricing.tiers · feature.grid · logo.cloud · social.testimonial

Geo & Maps (4)

geo.country_highlight · geo.route_draw · geo.map_zoom_pin · geo.globe_spin

Media & Imagery (4)

media.broll_card · media.video_card · media.annotated_photo · media.double_exposure

Quotes, Attribution & Lower Thirds (3)

quote.callout · source.card · name.lower_third

CTA / Engagement / YouTube / HUD (8)

cta.subscribe_button · cta.engagement_bar · poll.choice_card · hud.scoreboard · progress.chapter_bar · timer.countdown · youtube.info_card_teaser · youtube.end_screen_layout

FX / Ambient / Backgrounds (10)

fx.radial_pulse · fx.confetti_burst · fx.gradient_sweep · fx.color_blocks · fx.filter · fx.particles · fx.lens_flare · fx.audio_waveform · bg.gradient_morph · bg.blob_gradient

Glass / UI / Motion Devices (8)

layout.glass_card · layout.glass_menu · layout.social_notification · layout.ticker_tape · layout.composition · button.liquid_glass · button.glass_toggle · card.flip

Cinematic & Depth (2)

scene.parallax_3d · focus.pointer_highlight

Utility (5)

concept.icon_lottie · logo.reveal_stroke · privacy.blur_region · time.timeline_marker · none

---

Goals & briefs

The agent classifies every prompt into a production goal and selects the appropriate workflow:

Goals: viral social · marketing promo · sales demo · educational · raw-to-polished · event highlight · talking head · platform export · scene edits · conversational inquiry

Export styles: viral clips · single · multi-platform · none (scene-edit only)

For complex briefs, pass a structured prompt with multiple objectives. The DAG compiler expands them into ordered, gated tasks — so "reframe vertical, add captions, remove silences, and export for TikTok" runs as four verified steps, not four separate calls.

---

API reference

All paths are under /api/v1/misc/openclaw.

Each management group above is also exposed as a typed MCP tool (see Tools the AI sees).

LEVEA_API_URL is the bare host https://api.livecore.ai — not a full path, not studio.livecore.ai, and not the in-product /api/v1/misc/editor/ route.

Request

Every edit goes through one tool — autonomous_edit. Pass a natural-language description; the agent plans, executes, verifies, and exports. No tool list to memorize, no structured params to learn.

curl -sS -X POST "$LEVEA_API_URL/api/v1/misc/openclaw/v1/execute" \
  -H "Authorization: Bearer $LEVEA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "autonomous_edit",
    "params": {
      "prompt": "Make this a TikTok-ready viral clip: vertical reframe, bold captions, remove silences, motion-track the speaker, and export."
    },
    "project_id": "my-project"
  }'

prompt is the only required field. Everything else is optional:

Response

{
  "type": "success | partial_success",
  "tool": "autonomous_edit",
  "success": true,
  "partial": false,
  "status": "completed | failed | awaiting_approval",
  "scene": { },
  "reply": "Human-readable summary of what changed",
  "message": "Short status message",
  "videoUrl": "https://.../output.mp4",
  "jobId": "12345",
  "viral_clips": [ ],
  "zip_url": "https://.../clips.zip",
  "activeTasks": [ ],
  "pendingAsyncJobs": [ ],
  "verificationPassed": true,
  "verificationIssues": [],
  "quality": {
    "score": 0.94,
    "stepsTotal": 7,
    "stepsFailed": 0,
    "hasExportedMedia": true,
    "mediaReachable": true
  },
  "processingTime": 12.3,
  "workingMemory": { }
}

jobId is the numeric task_id (a string of digits), not a task_-prefixed value. quality is a calibrated post-run score block; partial is the top-level partial-success flag (mirrors type: "partial_success").

SSE streaming

Set Accept: text/event-stream or ?stream=true. Notable event types:

Async job lifecycle

Generation actions return immediately with a jobId. Poll until terminal:

JOB_ID=$(jq -r '.jobId // .activeTasks[0].intent.job_id' /tmp/result.json)
while true; do
  STATUS=$(curl -sS "$LEVEA_API_URL/api/v1/misc/openclaw/v1/jobs/$JOB_ID" \
    -H "Authorization: Bearer $LEVEA_API_KEY" | jq -r '.status')
  echo "Status: $STATUS"
  case "$STATUS" in completed|succeeded|failed|error|cancelled) break;; esac
  sleep 5
done

curl -sS "$LEVEA_API_URL/api/v1/misc/openclaw/v1/jobs/$JOB_ID" \
  -H "Authorization: Bearer $LEVEA_API_KEY" | jq '.result'

Auto-export

After any mutating tool call that changed the scene (including partial successes), if no export is already queued the route auto-fires one synchronously — so the response carries a real videoUrl. Read-only and conversational autonomous_edit calls do not trigger auto-export.

Plan approval

Pass requirePlanApproval: true to make the agent stop after planning. It returns status: "awaiting_approval" + populated workingMemory. Resume by sending the same workingMemory with an approval prompt (yes, approve, go, do it, confirm).

---

Who it's for

---

vs. legacy editors

---

Safety & limits

• Beta — outputs can be wrong. Preview every result before sharing. For irreversible workflows, pass requirePlanApproval: true to inspect and approve the plan first.
• API-key authentication on every request.
• Destructive actions (mass deletes, clears) require explicit confirmation params.
• Output is verified after execution; auto-repairs on failure.
• Concurrent identical requests are deduplicated server-side.
• Rate-limited per API key. Read-only ~1–3s · structural edits ~3–10s · async generation 30s–5min per artifact · viral-clip / multi-platform exports several minutes.

---

Supported media

---

Links & channels

Versions aren't pinned here — npx -y levea-mcp-server and the ClawHub listing always pull the latest.

Support — setup help, integration questions, or issue reports: brajendrak00068@gmail.com

License — MIT