Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
# OpenMontage â€" Skill Index
> For the full agent onboarding guide, see [`AGENT_GUIDE.md`](../AGENT_GUIDE.md) in the project root.
This file lists all available Layer 2 skills and documents the 3-layer knowledge architecture.
## Knowledge Architecture
```
Layer 1: tools/tool_registry.py "What tools exist and what they can do"
tools/base_tool.py Each tool declares: capabilities, tier, status,
dependencies, cost, and agent_skills[]
â†" agent_skills[] points to â†"
Layer 2: skills/ "How OpenMontage uses these tools"
Project-specific conventions: Pipeline integration, artifact mappings,
{core,creative,meta,pipelines}/ enhancement chain order, quality checklists
â†" references underlying tech in â†"
Layer 3: .agents/skills/ "How the technology itself works"
Generic API knowledge from Correct import paths, code patterns,
skills.sh (47 installed skills) constraints, parameters â€" tech-agnostic
```
**How the agent uses this:**
1. The orchestrator queries Layer 1 (`tool_registry.support_envelope()` ) to see what's available
2. Each tool's `agent_skills[]` field names the Layer 3 skills it relies on
3. Layer 2 skills (this directory) teach the agent OpenMontage-specific conventions
4. Layer 3 skills (`.agents/skills/` ) provide generic API knowledge, loaded on-demand
## Capability Families & Tool Discovery
Every tool declares a `capability` (what it does) and a `provider` (who/what powers it). The registry groups tools by capability so agents can discover all options for a given task.
### Selector / Provider Pattern
For capability families with multiple providers (TTS, video generation), the architecture uses:
- **Selector tool** (`tts_selector` , `video_selector` , `image_selector` ) — routes to the best available provider based on requirements, API key availability, and cost. Selectors auto-discover providers from the registry. Agents should default to selectors when the user hasn't specified a provider.
- **Provider tools** — call a specific provider directly. Agents use these when the user explicitly requests a provider or when the selector's routing isn't appropriate.
### Capability Family Reference
**Do not maintain a hardcoded tool list.** The registry is the single source of truth. Query it at runtime:
```bash
python -c "from tools.tool_registry import registry; import json; registry.discover(); print(json.dumps(registry.capability_catalog(), indent=2))"
```
Key capability families to look for in the output:
| Capability | Selector | Discovery |
|---|---|---|
| `tts` | `tts_selector` | Auto-discovers all `capability="tts"` tools |
| `video_generation` | `video_selector` | Auto-discovers all `capability="video_generation"` tools |
| `image_generation` | `image_selector` | Auto-discovers all `capability="image_generation"` tools |
| `audio_processing` | — | FFmpeg-based local tools |
| `enhancement` | — | Mixed providers |
| `analysis` | — | Mixed providers |
docs(prompting): adopt 5-aspect video specification across skills
Incorporate the structured taxonomy from Lin et al. "Building a Precise
Video Language with Human-AI Oversight" (CMU/Harvard, arXiv 2604.21718v2).
The paper proves prompts structured around five aspects (Subject /
Subject Motion / Scene / Spatial Framing / Camera) unlock controllable
cinematography in fine-tuned video generation models. Off-the-shelf VLMs
already nail subject and scene; the gains live in motion, spatial, and
camera, which prompts routinely omit.
Universal layer (skills/creative/video-gen-prompting.md, +125 lines):
- 5-aspect prompt skeleton replaces flat formula
- Camera movements regrouped (translation / rotation / lens-only) with
dolly!=zoom, pan!=truck, bird's-eye!=aerial disambiguations
- New primitive tables: camera height, camera angle, POV, lens
distortion (fisheye vs barrel), focus / DoF (rack / pull / tracking),
playback speed (6 modes), subject transitions
- Order-matters and self-contained-prompt rules
- Identity anchoring rule for multi-shot
- Strict static-shot rule, anti-subjective callout, overlays-not-depth
- Per-model word-count guidance
Per-model guides (sora, veo, hunyuan, ltx, seedance):
- Add the primitives each model honors literally
- Word-count sweet spots per model
- Strengthen seedance verbatim-identity and subject-transition guidance
Pipeline directors (cinematic / explainer / animation scene-director,
cinematic / explainer asset-director):
- 5-aspect scene-plan checklist (per-pipeline adapted)
- Overlays-not-depth callout
- Pre / critique / post self-review loop for generation prompts
Reviewer (skills/meta/reviewer.md):
- CHAI critique-quality rules: accurate / complete / constructive
- Critical findings now require a proposed_fix
Storytelling, cinematic, broll, video-reference-analyst:
- Anti-subjective rule (replace mood adjectives with visual causes)
- Camera-intent-per-beat for script writers
- POV column in stock-footage query templates
- 5-aspect structured output mandatory for reference-video analysis
skills/INDEX.md: video-gen-prompting marked as canonical 5-aspect spec.
2026-04-28 08:11:31 -07:00
| `character_animation` | — | Local character specs, SVG rigs, pose libraries, action timelines, previews, and QA |
2026-08-13 07:48:38 -07:00
| `3d_world_generation` | — | Local semantic terrain, procedural scattering, explicit landmarks, diagnostics, and HyperFrames/Three.js fly-through workspaces |
| `3d_asset_acquisition` | — | Rights-safe local GLTF/GLB catalogs and provenance |
| `3d_asset_generation` | — | Atlas/fal textured mesh generation and reconstruction for unique scene assets |
| `3d_world_rendering` | — | Blender assembly and production rendering of detailed worlds |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| `graphics` | — | Local rendering tools |
2026-08-03 02:14:01 -07:00
| `music_library` | — | Discovers user-provided local tracks |
| `music_search` | — | Discovers royalty-free search/download providers |
| `music_generation` | — | Discovers paid/local generation providers |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| `subtitle` | — | Pure Python |
| `avatar` | — | Local GPU models |
| `video_post` | — | FFmpeg-based local tools |
### Adding New Tools
1. Place the tool in the correct capability folder (or create a new one under `tools/` )
2. Set `capability` and `provider` in the class definition
3. If joining a multi-provider family, the existing selector discovers it automatically
4. Attach relevant Layer 2 and Layer 3 skills via `agent_skills[]`
5. The registry discovers tools automatically — no manual registration needed
6. **No other files need updating ** — selectors, manifests, and instructions all derive from the registry
## Core Skills
| Skill | File | Trigger | Agent Skills (Layer 3) |
|-------|------|---------|----------------------|
2026-06-25 19:00:47 +05:30
| FFmpeg | `core/ffmpeg.md` | Video encoding, filtering, composition | `ffmpeg` , `video-toolkit` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| Remotion | `core/remotion.md` | React-based composition, Phase 3+ | `remotion-best-practices` , `remotion` |
chore(deps,skills): bump Remotion 4.0.441→4.0.484; re-vendor HyperFrames skills at v0.7.17
## Remotion bump (mechanical, semver-safe)
remotion-composer: Remotion 4.0.441 → 4.0.484 (43 patch versions, all
within 4.0.x). Includes the seven core packages: remotion + @remotion/cli,
captions, google-fonts, media, player, transitions.
Smoke test: re-rendered the compound-snowball atelier composition through
the unchanged tool path; final_review status=pass, atelier checks clean.
Note: package.json also carries d3-geo@^3.1.1 forward — this line was
already in the working tree from prior unrelated WIP and is not part of
this bump. Removing it would leave package-lock mismatched, so it's
preserved as-is here; clean up separately when its consumer lands.
## HyperFrames skills re-vendor (0.4.2 → 0.7.17)
The runtime invoked by hyperframes_compose (`npx hyperframes`) was
already pulling 0.7.17 on every render, but the vendored skill docs the
agent reads were frozen at 0.4.2-era. This commit closes that gap.
Re-vendored from upstream commit 3351fb1a (tag v0.7.17, 2026-06-27):
Re-vendored core 4 (restructured upstream):
- hyperframes (slim entry; deep content moved to focused skills)
- hyperframes-cli (1 → 7 files; covers validate/inspect/snapshot/
benchmark/lambda natively, dropping the obsolete
OM-local validate patch)
- hyperframes-registry
- website-to-video (renamed upstream from website-to-hyperframes)
Newly vendored (8 strategic additions in 0.5–0.7):
- hyperframes-core composition contract (data-*/tracks/sub-comps)
- hyperframes-creative palette, type, narration, beat planning
- hyperframes-media TTS, BGM, SFX, transcription, captions, bg-remove
- hyperframes-animation all motion knowledge (rules, blueprints,
transitions, 7 runtime adapters)
- media-use agent Media OS (one `resolve` verb for
BGM/SFX/image/icon; project + global cache)
- motion-graphics short design-led motion patterns
- remotion-to-hyperframes migration guidance (directly relevant since
OpenMontage runs both runtimes)
- music-to-video beat-synced video using `hyperframes beats`
Intentionally NOT vendored (HF-workflow-specific; would compete with
OpenMontage pipeline routing): embedded-captions, faceless-explainer,
general-video, pr-to-video, product-launch-video, slideshow,
talking-head-recut. Re-evaluate per pipeline need.
PROVENANCE.md refreshed with the new vendor point + re-sync instructions.
## GSAP CDN pin
.agents/skills/hyperframes/SKILL.md: gsap@3.14.2 → gsap@3 (auto-latest 3.x
on jsdelivr; avoids future drift without breaking the API surface).
## Doctrine updates routing to new skill structure
- skills/INDEX.md — HyperFrames row expanded to enumerate the 12 vendored
skills and their roles.
- skills/meta/animation-runtime-selector.md — runtime decision matrix
updated for the rename (website-to-hyperframes → website-to-video) and
three new rows added: beat-synced music videos, Remotion→HF porting,
and the media-use resolve verb. The HyperFrames composition row in the
animation-library matrix split into four (core/creative/media/animation)
per the upstream skill structure.
- skills/core/hyperframes.md — Layer-2 routing skill rewritten to point at
the new focused skills and all website-to-hyperframes references renamed.
2026-06-27 14:00:00 -07:00
| HyperFrames | `core/hyperframes.md` | HTML/CSS/GSAP composition runtime — kinetic typography, music-to-video, product promos, website capture. Vendored at v0.7.17 (2026-06-27). | `hyperframes` (router) → `hyperframes-core` (contract), `hyperframes-creative` (palette/type/narration), `hyperframes-media` (TTS/BGM/SFX/captions), `hyperframes-animation` (all motion), `hyperframes-cli` , `hyperframes-registry` , `media-use` , `motion-graphics` , `music-to-video` (beats-driven), `website-to-video` , `remotion-to-hyperframes` (migration), `gsap-core` , `gsap-timeline` |
feat(stt): add Azure AI Speech as an optional cloud speech-to-text provider
Add an Azure AI Speech transcription tool. It is opt-in: when
AZURE_SPEECH_KEY is configured the agent may prefer it for cloud STT,
while the local faster-whisper `transcriber` stays the default offline
path. Shared pipeline manifests are intentionally left unchanged, so no
default provider selection is altered for existing users.
- tools/analysis/azure_stt.py: new `azure_stt` tool (capability=analysis,
provider=azure) calling the Fast Transcription REST API. The local file
is uploaded via multipart and transcribed synchronously with word-level
timestamps and optional diarization — no Blob storage or async polling.
Output schema mirrors `transcriber` exactly, so it is a drop-in for
`subtitle_gen` and other transcript consumers. Follows the existing
provider-tool conventions (env-var status check, `_transcribe` helper,
cost_usd/model on the result, fallback="transcriber").
- Auto-discovered by the registry; no registry or selector changes.
- tests/tools/test_azure_stt.py: contract, discovery, status, response
mapping, execute guardrails, and a mocked-network success path (no live
API calls).
- .agents/skills + .claude/skills: azure-speech-to-text Layer-3 skill.
- docs/PROVIDERS.md: Azure AI Speech setup, API notes, and pricing.
- .env.example, skills/INDEX.md, AGENT_GUIDE.md: document the optional
cloud STT path alongside the default whisper transcriber.
2026-07-10 23:14:14 +05:30
| WhisperX | `core/whisperx.md` | Transcription with word-level timestamps — default STT (offline, free) | `speech-to-text` |
| Azure STT | (tool: `azure_stt` ) | Optional cloud speech-to-text, word-level timestamps — preferred when `AZURE_SPEECH_KEY` is set | `azure-speech-to-text` |
feat(tts): add Azure AI Speech as an optional cloud text-to-speech provider
Neural TTS via the synchronous REST v1 endpoint (SSML body, no token
exchange or job polling). Shares one Speech resource with azure_stt —
AZURE_SPEECH_KEY + AZURE_SPEECH_REGION unlock both directions; optional
AZURE_TTS_ENDPOINT overrides the TTS host (a different subdomain than
the STT endpoint). piper_tts remains the default offline path.
- tools/audio/azure_tts.py: azure_tts tool (capability=tts), voice
shortlist aliases, SSML prosody/style, mp3/wav output, cost tracking
- tests/tools/test_azure_tts.py: contract, discovery, status, SSML,
and mocked execute tests (21 tests, no live network)
- .agents/.claude skills: azure-text-to-speech Layer-3 skill
- docs: PROVIDERS.md section + tables, ARCHITECTURE.md inventories,
AGENT_GUIDE.md + skills/INDEX.md rows, asset-director TTS cheatsheet,
.env.example
2026-07-13 20:04:28 +05:30
| Azure TTS | (tool: `azure_tts` ) | Optional cloud neural narration (SSML prosody, express-as styles) — same Speech key as `azure_stt` | `azure-text-to-speech` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| Subtitle Sync | `core/subtitle-sync.md` | Subtitle timing and alignment | `remotion-best-practices` |
| Color Grading | `core/color-grading.md` | FFmpeg color profiles, LUT workflow, accessibility | `ffmpeg` |
## Creative Skills
| Skill | File | Trigger | Agent Skills (Layer 3) |
|-------|------|---------|----------------------|
2026-06-25 19:00:47 +05:30
| Video Editing | `creative/video-editing.md` | Cut decisions, pacing, rhythm | `ffmpeg` , `video-toolkit` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| Enhancement Strategy | `creative/enhancement-strategy.md` | Overlay placement and density | `ffmpeg` |
| Data Visualization | `creative/data-visualization.md` | Chart type selection, animation, label placement | `d3-viz` , `remotion-best-practices` |
2026-06-25 19:00:47 +05:30
| Video Stitching | `creative/video-stitching.md` | Multi-clip assembly, AI clip chaining, spatial composition | `ffmpeg` , `video-toolkit` |
docs(prompting): adopt 5-aspect video specification across skills
Incorporate the structured taxonomy from Lin et al. "Building a Precise
Video Language with Human-AI Oversight" (CMU/Harvard, arXiv 2604.21718v2).
The paper proves prompts structured around five aspects (Subject /
Subject Motion / Scene / Spatial Framing / Camera) unlock controllable
cinematography in fine-tuned video generation models. Off-the-shelf VLMs
already nail subject and scene; the gains live in motion, spatial, and
camera, which prompts routinely omit.
Universal layer (skills/creative/video-gen-prompting.md, +125 lines):
- 5-aspect prompt skeleton replaces flat formula
- Camera movements regrouped (translation / rotation / lens-only) with
dolly!=zoom, pan!=truck, bird's-eye!=aerial disambiguations
- New primitive tables: camera height, camera angle, POV, lens
distortion (fisheye vs barrel), focus / DoF (rack / pull / tracking),
playback speed (6 modes), subject transitions
- Order-matters and self-contained-prompt rules
- Identity anchoring rule for multi-shot
- Strict static-shot rule, anti-subjective callout, overlays-not-depth
- Per-model word-count guidance
Per-model guides (sora, veo, hunyuan, ltx, seedance):
- Add the primitives each model honors literally
- Word-count sweet spots per model
- Strengthen seedance verbatim-identity and subject-transition guidance
Pipeline directors (cinematic / explainer / animation scene-director,
cinematic / explainer asset-director):
- 5-aspect scene-plan checklist (per-pipeline adapted)
- Overlays-not-depth callout
- Pre / critique / post self-review loop for generation prompts
Reviewer (skills/meta/reviewer.md):
- CHAI critique-quality rules: accurate / complete / constructive
- Critical findings now require a proposed_fix
Storytelling, cinematic, broll, video-reference-analyst:
- Anti-subjective rule (replace mood adjectives with visual causes)
- Camera-intent-per-beat for script writers
- POV column in stock-footage query templates
- 5-aspect structured output mandatory for reference-video analysis
skills/INDEX.md: video-gen-prompting marked as canonical 5-aspect spec.
2026-04-28 08:11:31 -07:00
| Video Gen Prompting | `creative/video-gen-prompting.md` | Universal video generation prompt vocabulary; **canonical 5-aspect spec ** (Subject / Motion / Scene / Spatial / Camera); ~200 cinematography primitives | `ai-video-gen` , `ltx2` , `create-video` |
video-gen: adopt Seedance 2.0 as preferred premium default
Seedance 2.0 is now routed as the top-ranked video generator whenever a
premium gateway is available. Touches the tool layer, scoring engine,
cinematic pipeline, and both skill layers so discovery works from every
entry point.
- tools/video/seedance_video: BETA stability, quality_score=0.95, add
reference_to_video operation plus 9 img + 3 vid + 3 audio ceilings,
fix pre-existing upload_image_fal import
- tools/base_tool: surface optional quality_score / success_rate /
latency fields in get_info so the scorer can read them
- lib/scoring: fix reliability enum-vs-string bug that was pinning every
available tool to 0.0, switch to overlap coefficient so rich best_for
descriptions aren't penalized, add premium-cinematic feature bonus
- pipeline_defs/cinematic + cinematic asset-director: add pixabay_music
and freesound_music, restore pixabay-first music default
- cinematic compose-director: mandatory Remotion preflight at stage entry
- New Layer 3 .agents/skills/seedance-2-0/SKILL.md (8-part prompt
structure, multi-shot, lip-sync, reference-to-video, provider landscape)
- New Layer 2 skills/creative/prompting/seedance-prompting.md
- Update ai-video-gen, video-gen-prompting, AGENT_GUIDE, INDEX to flag
Seedance 2.0 as the preferred premium default and make the skill
discoverable from every routing path
2026-04-17 21:51:54 -07:00
| ↳ Seedance Prompting | `creative/prompting/seedance-prompting.md` | **Preferred premium default. ** Seedance 2.0 8-component structure, multi-shot, lip-sync, reference-to-video | `seedance-2-0` , `ai-video-gen` |
2026-04-05 15:31:37 -07:00
| ↳ Grok Prompting | `creative/prompting/grok-prompting.md` | Grok image/video prompting, edit flows, reference-image video | `grok-media` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| ↳ Sora Prompting | `creative/prompting/sora-prompting.md` | Sora 2 structured template, advanced fields | `ai-video-gen` |
| ↳ VEO Prompting | `creative/prompting/veo-prompting.md` | VEO 3.1 14-component structure, art movements | `ai-video-gen` |
| ↳ LTX Prompting | `creative/prompting/ltx-prompting.md` | LTX-2 6-element structure, audio prompting | `ltx2` |
| ↳ HunyuanVideo Prompting | `creative/prompting/hunyuan-prompting.md` | HunyuanVideo formula, I2V best practices | â€" |
| Storytelling | `creative/storytelling.md` | Narrative structure, hooks, pacing, Mayer's principles | â€" |
| Sound Design | `creative/sound-design.md` | Audio ducking, LUFS targets, SFX timing, AI TTS mixing | `elevenlabs` |
| Typography | `creative/typography.md` | Font selection, text sizing, safe zones, caption styling | â€" |
| ManimCE Usage | `creative/manim-usage.md` | Scene composition, animation timing, color usage | `manimce-best-practices` |
| Image Gen Usage | `creative/image-gen-usage.md` | Prompt consistency, hero reference, batch strategy | `flux-best-practices` , `bfl-api` |
2026-04-05 15:31:37 -07:00
| Image Provider Usage | `creative/image-provider-usage.md` | Provider selection (FLUX/Grok/OpenAI/Recraft/stock), cost-quality tradeoffs | `flux-best-practices` , `bfl-api` , `grok-media` |
2026-08-13 07:48:38 -07:00
| 3D World Generation | `creative/3d-world-generation.md` | Semantic world planning, asset sourcing/generation, Blender assembly, and fidelity review | `3d-asset-generation` , `threejs-world-generation` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| B-Roll Planning | `creative/broll-planning.md` | Stock vs. generated decision, query construction, footage evaluation | — |
| Stock Sourcing Usage | `creative/stock-sourcing-usage.md` | Pexels/Pixabay usage, parameters, licensing, integration | — |
| Scene Detect Usage | `creative/scene-detect-usage.md` | Threshold tuning, algorithm selection, content presets | â€" |
| Diagram Gen Usage | `creative/diagram-gen-usage.md` | Complexity limits, progressive building, themes | `beautiful-mermaid` |
| Music Gen Usage | `creative/music-gen-usage.md` | BPM selection, prompt engineering, duration matching | `music` , `elevenlabs` |
| Background Removal | `creative/bg-remove-usage.md` | Model selection, alpha matting, compositing workflows | â€" |
| Upscaling | `creative/upscale-usage.md` | Scale factor, model selection, face-aware upscaling | â€" |
| Face Restoration | `creative/face-restore-usage.md` | CodeFormer/GFPGAN selection, fidelity tuning, vs face_enhance | â€" |
| Lip Sync | `creative/lip-sync-usage.md` | Wav2Lip model selection, dubbing workflows, input requirements | `faceswap` |
| Talking Head Gen | `creative/talking-head-gen-usage.md` | SadTalker/MuseTalk, photo-to-video, expression tuning | `avatar-video` |
| Video Understanding | `creative/video-understand-usage.md` | Visual QA, quality gating, scene classification | `video-understand` |
## Pipeline Type Skills
Pipeline type skills provide production guidance for specific video formats, independent of the animated-explainer or talking-head pipeline.
| Skill | File | When to Use |
|-------|------|-------------|
| Short-Form | `creative/short-form.md` | TikTok, Reels, Shorts â€" vertical 9:16, under 60s |
| Long-Form | `creative/long-form.md` | YouTube 10+ min â€" chapters, retention, end screens |
| Screen Recording | `creative/screen-recording.md` | Code walkthroughs, tutorials, software demos |
| Animation Pipeline | `creative/animation-pipeline.md` | Motion graphics, easing, transitions, composition |
2026-08-13 07:48:38 -07:00
| 3D World Generation | `creative/3d-world-generation.md` | Continuous Three.js terrain worlds with semantic regions, explicit blockout/production tiers, licensed GLTF/PBR assets, diagnostics, and deterministic camera paths |
docs(prompting): adopt 5-aspect video specification across skills
Incorporate the structured taxonomy from Lin et al. "Building a Precise
Video Language with Human-AI Oversight" (CMU/Harvard, arXiv 2604.21718v2).
The paper proves prompts structured around five aspects (Subject /
Subject Motion / Scene / Spatial Framing / Camera) unlock controllable
cinematography in fine-tuned video generation models. Off-the-shelf VLMs
already nail subject and scene; the gains live in motion, spatial, and
camera, which prompts routinely omit.
Universal layer (skills/creative/video-gen-prompting.md, +125 lines):
- 5-aspect prompt skeleton replaces flat formula
- Camera movements regrouped (translation / rotation / lens-only) with
dolly!=zoom, pan!=truck, bird's-eye!=aerial disambiguations
- New primitive tables: camera height, camera angle, POV, lens
distortion (fisheye vs barrel), focus / DoF (rack / pull / tracking),
playback speed (6 modes), subject transitions
- Order-matters and self-contained-prompt rules
- Identity anchoring rule for multi-shot
- Strict static-shot rule, anti-subjective callout, overlays-not-depth
- Per-model word-count guidance
Per-model guides (sora, veo, hunyuan, ltx, seedance):
- Add the primitives each model honors literally
- Word-count sweet spots per model
- Strengthen seedance verbatim-identity and subject-transition guidance
Pipeline directors (cinematic / explainer / animation scene-director,
cinematic / explainer asset-director):
- 5-aspect scene-plan checklist (per-pipeline adapted)
- Overlays-not-depth callout
- Pre / critique / post self-review loop for generation prompts
Reviewer (skills/meta/reviewer.md):
- CHAI critique-quality rules: accurate / complete / constructive
- Critical findings now require a proposed_fix
Storytelling, cinematic, broll, video-reference-analyst:
- Anti-subjective rule (replace mood adjectives with visual causes)
- Camera-intent-per-beat for script writers
- POV column in stock-footage query templates
- 5-aspect structured output mandatory for reference-video analysis
skills/INDEX.md: video-gen-prompting marked as canonical 5-aspect spec.
2026-04-28 08:11:31 -07:00
| Character Animation Pipeline | `pipelines/character-animation/` | Rigged local cartoon characters, pose libraries, action timelines, SVG/Canvas/Remotion/HyperFrames rendering |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| Cinematic | `creative/cinematic.md` | Letterbox, film pacing, layered audio, color grading |
## Pipeline Stage Director Skills
Stage director skills teach the agent HOW to execute each pipeline stage. Each skill is a detailed markdown file with process steps, quality rubrics, and self-evaluation criteria.
### Animated Explainer Pipeline (`pipelines/explainer/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/explainer/executive-producer.md` | `all` | **8-stage serial orchestration, quality gates, cross-stage checks, send-back ** |
| **Research Director ** | `pipelines/explainer/research-director.md` | `research` | **Web research methodology, 5 search batches, landscape/trending/data/audience/expert analysis ** |
| **Proposal Director ** | `pipelines/explainer/proposal-director.md` | `proposal` | **Concept options from research, production plan, cost estimate, approval gate ** |
| Script Director | `pipelines/explainer/script-director.md` | `script` | Narrative architecture, timing, enhancement cues, research integration |
| Scene Director | `pipelines/explainer/scene-director.md` | `scene_plan` | Visual planning, technique library, feasibility |
| Asset Director | `pipelines/explainer/asset-director.md` | `assets` | TTS, image gen, diagram gen, music, budget |
| Edit Director | `pipelines/explainer/edit-director.md` | `edit` | Timeline assembly, subtitles, audio ducking |
| Compose Director | `pipelines/explainer/compose-director.md` | `compose` | FFmpeg/Remotion render, audio mixing |
| Publish Director | `pipelines/explainer/publish-director.md` | `publish` | SEO metadata, chapters, export packaging |
> **Note:** The old `idea-director.md` still exists for reference but is superseded by the research + proposal two-stage flow in v2.0. The talking-head pipeline continues to use its own `idea-director`.
### Talking Head Pipeline (`pipelines/talking-head/`)
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| Idea Director | `pipelines/talking-head/idea-director.md` | `idea` | Footage inspection, content assessment |
| Script Director | `pipelines/talking-head/script-director.md` | `script` | Transcription, section segmentation |
| Scene Director | `pipelines/talking-head/scene-director.md` | `scene_plan` | Enhancement planning, overlay placement |
| Asset Director | `pipelines/talking-head/asset-director.md` | `assets` | Subtitle gen, audio extraction |
| Edit Director | `pipelines/talking-head/edit-director.md` | `edit` | Cut assembly, subtitle config |
| Compose Director | `pipelines/talking-head/compose-director.md` | `compose` | Enhancement chain, render |
| Publish Director | `pipelines/talking-head/publish-director.md` | `publish` | Metadata, export packaging |
### Screen Demo Pipeline (`pipelines/screen-demo/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/screen-demo/executive-producer.md` | `all` | **7-stage serial orchestration, legibility gates, audio clarity, pacing checks ** |
| Idea Director | `pipelines/screen-demo/idea-director.md` | `idea` | Workflow scoping, UI density assessment, output-shape choice |
| Script Director | `pipelines/screen-demo/script-director.md` | `script` | Action mapping, procedural narration, speed planning |
| Scene Director | `pipelines/screen-demo/scene-director.md` | `scene_plan` | Crop planning, callout restraint, aspect-ratio viability |
| Asset Director | `pipelines/screen-demo/asset-director.md` | `assets` | Subtitle-first asset kit, audio cleanup, reusable overlays |
| Edit Director | `pipelines/screen-demo/edit-director.md` | `edit` | Tight timeline planning, speed notes, subtitle zone control |
| Compose Director | `pipelines/screen-demo/compose-director.md` | `compose` | Legibility-first render, crisp screen output, verification |
| Publish Director | `pipelines/screen-demo/publish-director.md` | `publish` | Searchable metadata, chapter packaging, thumbnail concepts |
### Clip Factory Pipeline (`pipelines/clip-factory/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/clip-factory/executive-producer.md` | `all` | **7-stage serial orchestration, clip selection gates, batch consistency, hook placement ** |
| Idea Director | `pipelines/clip-factory/idea-director.md` | `idea` | Batch strategy, clip families, yield planning |
| Script Director | `pipelines/clip-factory/script-director.md` | `script` | Transcript mining, ranking, standalone validation |
| Scene Director | `pipelines/clip-factory/scene-director.md` | `scene_plan` | Platform framing, safe zones, crop-viability planning |
| Asset Director | `pipelines/clip-factory/asset-director.md` | `assets` | Shared brand kit, rebased subtitles, batch audio consistency |
| Edit Director | `pipelines/clip-factory/edit-director.md` | `edit` | Hook-first mini-edits, series consistency |
| Compose Director | `pipelines/clip-factory/compose-director.md` | `compose` | Multi-job rendering, batch resilience, per-output verification |
| Publish Director | `pipelines/clip-factory/publish-director.md` | `publish` | Posting order, platform copy, batch cataloging |
### Podcast Repurpose Pipeline (`pipelines/podcast-repurpose/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/podcast-repurpose/executive-producer.md` | `all` | **7-stage serial orchestration, audio preservation gates, clip quality, multi-deliverable ** |
| Idea Director | `pipelines/podcast-repurpose/idea-director.md` | `idea` | Deliverable mix by source mode, realistic long-form planning |
| Script Director | `pipelines/podcast-repurpose/script-director.md` | `script` | Diarized transcript truth, highlight ranking, chapter mapping |
| Scene Director | `pipelines/podcast-repurpose/scene-director.md` | `scene_plan` | Source-faithful treatments, audiogram vs quote vs companion planning |
| Asset Director | `pipelines/podcast-repurpose/asset-director.md` | `assets` | Subtitle-first packaging, speaker assets, optional topic art |
| Edit Director | `pipelines/podcast-repurpose/edit-director.md` | `edit` | Hook-led podcast clips, quote hold time, companion simplicity |
| Compose Director | `pipelines/podcast-repurpose/compose-director.md` | `compose` | Audio-first rendering, deliverable prioritization |
| Publish Director | `pipelines/podcast-repurpose/publish-director.md` | `publish` | Episode cross-linking, guest attribution, staggered release logic |
### Cinematic Pipeline (`pipelines/cinematic/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/cinematic/executive-producer.md` | `all` | **7-stage serial orchestration, emotional pacing gates, color consistency, audio dynamics ** |
| Idea Director | `pipelines/cinematic/idea-director.md` | `idea` | Emotional arc selection, source truth, delivery-shape planning |
| Script Director | `pipelines/cinematic/script-director.md` | `script` | Beat mapping, dialogue selects, title-card restraint |
| Scene Director | `pipelines/cinematic/scene-director.md` | `scene_plan` | Hero-frame planning, reveal structure, transition limits |
| Asset Director | `pipelines/cinematic/asset-director.md` | `assets` | Source selects, support-insert discipline, music/ambience planning |
| Edit Director | `pipelines/cinematic/edit-director.md` | `edit` | Emotion-first pacing, reveal timing, audio-driven rhythm |
| Compose Director | `pipelines/cinematic/compose-director.md` | `compose` | Grade and mix finishing, frame-treatment judgment |
| Publish Director | `pipelines/cinematic/publish-director.md` | `publish` | Hero vs teaser packaging, poster-frame concepts |
### Animation Pipeline (`pipelines/animation/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/animation/executive-producer.md` | `all` | **8-stage serial orchestration, quality gates, motion consistency, math accuracy checks ** |
| **Research Director ** | `pipelines/animation/research-director.md` | `research` | **Topic + animation technique research, visual reference scan, mode-informed angles ** |
| **Proposal Director ** | `pipelines/animation/proposal-director.md` | `proposal` | **Animation mode selection (Manim/Remotion/AI/diagram), reuse strategy, cost estimate, approval gate ** |
| Script Director | `pipelines/animation/script-director.md` | `script` | Animation-ready beats, text restraint, research integration, mode-aware writing |
| Scene Director | `pipelines/animation/scene-director.md` | `scene_plan` | Animatic planning, transition systems, tool-path mapping |
| Asset Director | `pipelines/animation/asset-director.md` | `assets` | Deterministic asset choice, reusable motifs, feasibility truth |
| Edit Director | `pipelines/animation/edit-director.md` | `edit` | Hold timing, stagger rules, readable motion planning |
| Compose Director | `pipelines/animation/compose-director.md` | `compose` | Sharp render output, timing integrity, safe-zone checks |
| Publish Director | `pipelines/animation/publish-director.md` | `publish` | Animation-mode packaging, thumbnail-system alignment |
> **Note:** The old `idea-director.md` still exists for reference but is superseded by the research + proposal two-stage flow in v2.0.
### Hybrid Pipeline (`pipelines/hybrid/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/hybrid/executive-producer.md` | `all` | **7-stage serial orchestration, source/support balance gates, overlay density, coherence ** |
| Idea Director | `pipelines/hybrid/idea-director.md` | `idea` | Anchor-medium selection, support-layer planning, fallback visibility |
| Script Director | `pipelines/hybrid/script-director.md` | `script` | Source-vs-support beat mapping, dialogue retention, support justification |
| Scene Director | `pipelines/hybrid/scene-director.md` | `scene_plan` | Source-primary layout rules, overlay density control, variant-safe planning |
| Asset Director | `pipelines/hybrid/asset-director.md` | `assets` | Shared support kits, source-vs-generated asset tracking |
| Edit Director | `pipelines/hybrid/edit-director.md` | `edit` | Anchor-cut-first workflow, layered support timing, readable variants |
| Compose Director | `pipelines/hybrid/compose-director.md` | `compose` | Source/support balance checks, variant verification, coherent mix |
| Publish Director | `pipelines/hybrid/publish-director.md` | `publish` | Master-vs-derivative packaging, source-mix metadata |
### Avatar Spokesperson Pipeline (`pipelines/avatar-spokesperson/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/avatar-spokesperson/executive-producer.md` | `all` | **7-stage serial orchestration, lip-sync quality gates, presenter framing, CTA landing ** |
| Idea Director | `pipelines/avatar-spokesperson/idea-director.md` | `idea` | Avatar-path classification, CTA scoping, capability truth |
| Script Director | `pipelines/avatar-spokesperson/script-director.md` | `script` | Spoken-copy shaping, scene-safe pacing, text restraint |
| Scene Director | `pipelines/avatar-spokesperson/scene-director.md` | `scene_plan` | Presenter layout, background discipline, variant realism |
| Asset Director | `pipelines/avatar-spokesperson/asset-director.md` | `assets` | Avatar-path locking, narration resolution, minimal support kits |
| Edit Director | `pipelines/avatar-spokesperson/edit-director.md` | `edit` | Presenter-first cut planning, overlay timing, CTA landing |
| Compose Director | `pipelines/avatar-spokesperson/compose-director.md` | `compose` | Lip-sync verification, subtitle-safe framing, clean render checks |
| Publish Director | `pipelines/avatar-spokesperson/publish-director.md` | `publish` | Audience-led packaging, presenter-first thumbnail concepts |
### Localization Dub Pipeline (`pipelines/localization-dub/`) — v2.0
| Skill | File | Stage | Key Capabilities |
|-------|------|-------|-----------------|
| **Executive Producer ** | `pipelines/localization-dub/executive-producer.md` | `all` | **7-stage serial orchestration, translation accuracy gates, timing preservation, per-locale QA ** |
| Idea Director | `pipelines/localization-dub/idea-director.md` | `idea` | Scope definition, locale planning, glossary and review capture |
| Script Director | `pipelines/localization-dub/script-director.md` | `script` | Transcript truth, translated script packaging, term preservation |
| Scene Director | `pipelines/localization-dub/scene-director.md` | `scene_plan` | Dub-mode selection, timing-risk mapping, on-screen text planning |
| Asset Director | `pipelines/localization-dub/asset-director.md` | `assets` | Subtitle-first localization kit, dubbed audio generation, optional lip sync |
| Edit Director | `pipelines/localization-dub/edit-director.md` | `edit` | Locale-specific timelines, coverage planning, timing adjustments |
| Compose Director | `pipelines/localization-dub/compose-director.md` | `compose` | Per-locale rendering, subtitle-fit checks, output labeling |
| Publish Director | `pipelines/localization-dub/publish-director.md` | `publish` | Locale packaging, metadata precision, QA-note retention |
## Meta Skills
Cross-cutting skills that apply to all pipelines:
| Skill | File | Purpose |
|-------|------|---------|
2026-03-30 13:23:28 -07:00
| Onboarding | `meta/onboarding.md` | First-interaction greeting, capability discovery, starter prompts |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| Reviewer | `meta/reviewer.md` | Self-review protocol after every stage |
| Checkpoint Protocol | `meta/checkpoint-protocol.md` | When/how to checkpoint and request human approval |
| Skill Creator | `meta/skill-creator.md` | Dynamically create new skills during pipeline runs |
2026-06-27 10:32:07 -07:00
| Animation Runtime Selector | `meta/animation-runtime-selector.md` | Choose render runtime + animation library per scene |
2026-07-03 18:52:33 -07:00
| Taste Direction | `meta/taste-direction.md` | Convert a brief into taste dials, anti-patterns, and reference strategy for proposal/playbook/atelier work |
2026-06-27 10:32:07 -07:00
| Bespoke Composition (Atelier) | `meta/bespoke-composition.md` | Hand-author a composition from scratch (hero work) — no stock scene-types; routes art-direction → motion principles → engine mechanics → atelier render |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
## Style Playbooks
Style playbooks (`styles/*.yaml` ) define visual language, typography, motion, audio, and asset generation constraints. They are validated against `schemas/styles/playbook.schema.json` .
| Playbook | Category | Mood | Best For |
|----------|----------|------|----------|
| `clean-professional` | motion-graphics | polished, trustworthy | Corporate, educational, SaaS |
2026-07-03 18:52:33 -07:00
| `premium-minimalist` | minimalist | calm, editorial | Investor updates, expert explainers, product narratives |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| `flat-motion-graphics` | motion-graphics | energetic, bold | Social media, TikTok, startups |
| `minimalist-diagram` | whiteboard | focused, technical | Technical deep-dives, architecture |
Load via `styles/playbook_loader.py` : `load_playbook("clean-professional")`
## Installed Agent Skills (Layer 3)
All agent skills live in `.agents/skills/` and are managed via `npx skills add` .
Claude Code accesses them via symlinks in `.claude/skills/` .
| Category | Installed Skills | Source |
|----------|-----------------|--------|
chore(deps,skills): bump Remotion 4.0.441→4.0.484; re-vendor HyperFrames skills at v0.7.17
## Remotion bump (mechanical, semver-safe)
remotion-composer: Remotion 4.0.441 → 4.0.484 (43 patch versions, all
within 4.0.x). Includes the seven core packages: remotion + @remotion/cli,
captions, google-fonts, media, player, transitions.
Smoke test: re-rendered the compound-snowball atelier composition through
the unchanged tool path; final_review status=pass, atelier checks clean.
Note: package.json also carries d3-geo@^3.1.1 forward — this line was
already in the working tree from prior unrelated WIP and is not part of
this bump. Removing it would leave package-lock mismatched, so it's
preserved as-is here; clean up separately when its consumer lands.
## HyperFrames skills re-vendor (0.4.2 → 0.7.17)
The runtime invoked by hyperframes_compose (`npx hyperframes`) was
already pulling 0.7.17 on every render, but the vendored skill docs the
agent reads were frozen at 0.4.2-era. This commit closes that gap.
Re-vendored from upstream commit 3351fb1a (tag v0.7.17, 2026-06-27):
Re-vendored core 4 (restructured upstream):
- hyperframes (slim entry; deep content moved to focused skills)
- hyperframes-cli (1 → 7 files; covers validate/inspect/snapshot/
benchmark/lambda natively, dropping the obsolete
OM-local validate patch)
- hyperframes-registry
- website-to-video (renamed upstream from website-to-hyperframes)
Newly vendored (8 strategic additions in 0.5–0.7):
- hyperframes-core composition contract (data-*/tracks/sub-comps)
- hyperframes-creative palette, type, narration, beat planning
- hyperframes-media TTS, BGM, SFX, transcription, captions, bg-remove
- hyperframes-animation all motion knowledge (rules, blueprints,
transitions, 7 runtime adapters)
- media-use agent Media OS (one `resolve` verb for
BGM/SFX/image/icon; project + global cache)
- motion-graphics short design-led motion patterns
- remotion-to-hyperframes migration guidance (directly relevant since
OpenMontage runs both runtimes)
- music-to-video beat-synced video using `hyperframes beats`
Intentionally NOT vendored (HF-workflow-specific; would compete with
OpenMontage pipeline routing): embedded-captions, faceless-explainer,
general-video, pr-to-video, product-launch-video, slideshow,
talking-head-recut. Re-evaluate per pipeline need.
PROVENANCE.md refreshed with the new vendor point + re-sync instructions.
## GSAP CDN pin
.agents/skills/hyperframes/SKILL.md: gsap@3.14.2 → gsap@3 (auto-latest 3.x
on jsdelivr; avoids future drift without breaking the API surface).
## Doctrine updates routing to new skill structure
- skills/INDEX.md — HyperFrames row expanded to enumerate the 12 vendored
skills and their roles.
- skills/meta/animation-runtime-selector.md — runtime decision matrix
updated for the rename (website-to-hyperframes → website-to-video) and
three new rows added: beat-synced music videos, Remotion→HF porting,
and the media-use resolve verb. The HyperFrames composition row in the
animation-library matrix split into four (core/creative/media/animation)
per the upstream skill structure.
- skills/core/hyperframes.md — Layer-2 routing skill rewritten to point at
the new focused skills and all website-to-hyperframes references renamed.
2026-06-27 14:00:00 -07:00
| **Video Composition ** | `remotion-best-practices` , `remotion` , `hyperframes` (router), `hyperframes-core` , `hyperframes-creative` , `hyperframes-media` , `hyperframes-animation` , `hyperframes-cli` , `hyperframes-registry` , `media-use` , `motion-graphics` , `music-to-video` , `remotion-to-hyperframes` , `website-to-video` | `remotion-dev/skills` , `digitalsamba/claude-code-video-toolkit` , `heygen-com/hyperframes` (vendored v0.7.17, see `.agents/skills/hyperframes/PROVENANCE.md` ) |
2026-06-25 19:00:47 +05:30
| **Video Processing ** | `ffmpeg` , `video-toolkit` | `digitalsamba/claude-code-video-toolkit` |
2026-07-21 21:12:45 +09:00
| **TTS & Audio ** | `text-to-speech` , `speech-to-text` (whisper, default STT), `azure-speech-to-text` (optional cloud STT), `music` , `sound-effects` , `elevenlabs` , `fish-audio-tts` , `agents` , `setup-api-key` | `elevenlabs/skills` , `digitalsamba/claude-code-video-toolkit` , local OpenMontage skill |
2026-04-05 15:31:37 -07:00
| **Image Generation ** | `flux-best-practices` , `bfl-api` , `grok-media` | `black-forest-labs/skills` , local OpenMontage skill |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| **Math Animation ** | `manimce-best-practices` , `manimgl-best-practices` , `manim-composer` | `adithya-s-k/manim_skill` |
2026-08-13 07:48:38 -07:00
| **3D Graphics ** | `threejs-world-generation` (OpenMontage semantic-world workflow), `threejs-animation` , `threejs-fundamentals` , `threejs-geometry` , `threejs-interaction` , `threejs-lighting` , `threejs-loaders` , `threejs-materials` , `threejs-postprocessing` , `threejs-shaders` , `threejs-textures` | Local OpenMontage skill + `cloudai-x/threejs-skills` |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| **Diagrams ** | `beautiful-mermaid` , `d3-viz` | `intellectronica/agent-skills` , `davila7/claude-code-templates` |
| **Animation ** | `framer-motion` , `lottie-bodymovin` | `pproenca/dot-skills` , `dylantarre/animation-principles` |
| **Design ** | `tailwind-design-system` , `web-design-guidelines` , `vercel-react-best-practices` , `vercel-composition-patterns` | `wshobson/agents` , `vercel-labs/agent-skills` |
| **AI Video (HeyGen) ** | `heygen` , `avatar-video` , `create-video` , `faceswap` , `ai-video-gen` , `video-download` , `video-edit` , `video-translate` , `video-understand` , `visual-style` | `heygen-com/skills` |
2026-07-07 14:56:40 +08:00
| **AI Video/Image/TTS/Avatar (Kling Official) ** | `kling-official` - official direct API auth, Classic/Turbo/Omni task protocols, multi-reference Omni syntax, internal Elements/Account Usage helpers, callback notes, TTS voice parameters, avatar/lip-sync face selection, error handling, and cost governance for `kling_official_video` / `kling_official_image` / `kling_tts` / `kling_avatar` / `kling_lip_sync` | Local OpenMontage skill |
video-gen: adopt Seedance 2.0 as preferred premium default
Seedance 2.0 is now routed as the top-ranked video generator whenever a
premium gateway is available. Touches the tool layer, scoring engine,
cinematic pipeline, and both skill layers so discovery works from every
entry point.
- tools/video/seedance_video: BETA stability, quality_score=0.95, add
reference_to_video operation plus 9 img + 3 vid + 3 audio ceilings,
fix pre-existing upload_image_fal import
- tools/base_tool: surface optional quality_score / success_rate /
latency fields in get_info so the scorer can read them
- lib/scoring: fix reliability enum-vs-string bug that was pinning every
available tool to 0.0, switch to overlap coefficient so rich best_for
descriptions aren't penalized, add premium-cinematic feature bonus
- pipeline_defs/cinematic + cinematic asset-director: add pixabay_music
and freesound_music, restore pixabay-first music default
- cinematic compose-director: mandatory Remotion preflight at stage entry
- New Layer 3 .agents/skills/seedance-2-0/SKILL.md (8-part prompt
structure, multi-shot, lip-sync, reference-to-video, provider landscape)
- New Layer 2 skills/creative/prompting/seedance-prompting.md
- Update ai-video-gen, video-gen-prompting, AGENT_GUIDE, INDEX to flag
Seedance 2.0 as the preferred premium default and make the skill
discoverable from every routing path
2026-04-17 21:51:54 -07:00
| **AI Video (Premium) ** | `seedance-2-0` — preferred premium default (cinematic, trailer, multi-shot, lip-sync, synced audio); accessed via `seedance_video` (fal.ai) or `heygen_video` Avatar Shots | Local OpenMontage skill |
Initial release — OpenMontage: the first open-source agentic video production system
11 production pipelines, 47 tools, 124 agent skills.
Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and
free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
| **Infrastructure ** | `acestep` , `ltx2` , `playwright-recording` | `digitalsamba/claude-code-video-toolkit` |
