Tencent SkillHub · Developer Tools

Directoryahu

End-to-end pipeline for creating faceless Islamic story TikTok videos. Orchestrates multiple specialized agents: story research, scriptwriting, image generat...

skill openclawclawhub Free

0 Downloads

0 Stars

0 Installs

0 Score

High Signal

End-to-end pipeline for creating faceless Islamic story TikTok videos. Orchestrates multiple specialized agents: story research, scriptwriting, image generat...

⬇ 0 downloads ★ 0 stars Unverified but indexed

Install for OpenClaw

Quick setup

Download the package from Yavira.
Extract the archive and review SKILL.md first.
Import or place the package into your OpenClaw setup.

Requirements

Target platform: OpenClaw
Install method: Manual import
Extraction: Extract archive
Prerequisites: OpenClaw
Primary doc: SKILL.md

Package facts

Download mode: Yavira redirect
Package format: ZIP package
Source platform: Tencent SkillHub
What's included: visual_style_guide.md, story_agent.md, visual_agent.md, assembly_agent.md, global_config.json, orchestrator.py

Validation

Use the Yavira download entry.
Review SKILL.md after the package is downloaded.
Confirm the extracted package contains the expected setup assets.

Install with your agent

Agent handoff

Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.

Download the package from Yavira.
Extract it into a folder your agent can access.
Paste one of the prompts below and point your agent at the extracted folder.

New install

I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete.

Upgrade existing

I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run.

Open Send to Agent page Open JSON manifest Open Markdown brief

Trust & source

Release facts

Source: Tencent SkillHub
Verification: Indexed source record
Version: 1.0.0

Provenance

Publisher: mohamedzeidan2021
Source page: View original listing
Canonical URL: Open canonical page

Documentation

ClawHub primary doc Primary doc: SKILL.md 17 sections Open source page

How This Skill System Works

This is a multi-agent pipeline. There is ONE top-level orchestrator agent and FIVE specialist agents beneath it. Here's how they connect: ┌─────────────────────────────────────────────────────┐ │ ORCHESTRATOR AGENT │ │ (You feed THIS file — SKILL.md — to this agent) │ │ │ │ This agent reads SKILL.md, understands the full │ │ pipeline, and delegates to specialist agents. │ │ It is the "director" — it calls each agent in │ │ order, passes outputs between them, and handles │ │ errors. │ └──────────┬──────────────────────────────────────────┘ │ delegates to: │ ┌──────┴──────┐ │ │ ▼ ▼ ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐ │ Story │→ │ Script │→ │ Visual │ │ Voice │→ │Assembly│ │ Agent │ │ Agent │ │ Agent │ │ Agent │ │ Agent │ │ │ │ │ │ ↓ │ │ ↓ │ │ │ │ skill: │ │ skill: │ │(parallel)─(parallel)│ │ skill: │ │story.md│ │script │ │ skill: │ │ skill: │ │assembly│ │ │ │ .md │ │visual │ │voice.md│ │ .md │ │ │ │ │ │ .md │ │ │ │ │ └────────┘ └────────┘ └────────┘ └────────┘ └────────┘

How to set this up in OpenClaw:

Create the Orchestrator Agent Attach THIS file (SKILL.md) as its skill This agent needs the ability to call/invoke other agents It reads the pipeline flow below and executes step by step Create each Specialist Agent Each gets its own skill file from the agents/ folder Story Agent → agents/story_agent.md Script Agent → agents/script_agent.md Visual Agent → agents/visual_agent.md Voice Agent → agents/voice_agent.md Assembly Agent → agents/assembly_agent.md Attach tools to each agent Story Agent: web search (for hadith verification), file read Script Agent: no external tools needed (pure LLM reasoning) Visual Agent: image generation API (Flux/SDXL), face detection tool Voice Agent: ElevenLabs / OpenAI TTS API Assembly Agent: FFmpeg, file system access The Orchestrator calls agents in sequence: orchestrator receives: "Make a video about Prophet Nuh" → calls Story Agent → gets story JSON → calls Script Agent with story JSON → gets script JSON → calls Visual Agent with script JSON → gets image paths (parallel) → calls Voice Agent with script JSON → gets audio paths (parallel) → calls Assembly Agent with images + audio + script → gets final video

Visual Approach: Narrator + Story Scenes

Each video uses TWO types of visuals:

1. Narrator Scenes (brand anchor — same across ALL videos)

The faceless man in traditional Arabic clothing. Used for: Opening shot (first 2-3 seconds) — establishes the "host" Closing shot (last 2-3 seconds) — delivers the lesson / CTA The narrator character is your brand identity. Viewers recognize him across all your videos. He always looks the same: White thobe, red-checkered keffiyeh NEVER shows face (back shots, silhouettes, over-shoulder, hands) Same painterly art style, same color palette

2. Story Scenes (UNIQUE per video)

The middle 80% of each video shows what is actually happening in the story. These are completely unique to each video. Example — Story of Prophet Nuh (Noah): BeatVisualNuh preachingWide shot: lone figure on raised ground addressing a crowd in ancient city, people turning awayBuilding the arkClose-up: weathered hands hammering wood, massive wooden frame in backgroundThe mockeryCrowd of silhouettes laughing and pointing at the ship in the desertThe flood beginsDark storm clouds, rain hammering earth, water rising rapidlyAnimals boardingPairs of animal silhouettes walking toward a massive wooden ship at dawnThe floodEnormous waves, the ark riding the storm, lightning illuminating the sceneWaters recedeArk resting on a mountaintop, olive branch, golden light breaking through Example — Story of Prophet Yusuf (Joseph): BeatVisualThe dreamA boy looking up at night sky with 11 stars, sun, and moon arranged in a patternThe wellDark stone well in desert, rope descending into darknessThe caravanCamels in a line crossing desert dunes, golden dust in airThe palaceOrnate Egyptian palace interior, golden columns, silk drapesPrisonDim stone cell, single beam of light through a high windowThe reunionTwo silhouetted figures embracing in a field at sunset

Critical Rules for Story Scenes:

NO faces on any Prophets, Angels, or Sahaba — same faceless techniques Story scenes CAN show: landscapes, architecture, objects, animals, weather, hands/feet, silhouettes, wide shots where figures are tiny Art style must stay consistent WITHIN a single video Story scenes should match the emotional mood of the narration beat

How the Script Agent marks scene types:

{ "scenes": [ { "scene_number": 1, "scene_category": "narrator_opening", "narration_text": "What happens when an entire world turns against one man?", "visual_direction": { "description": "Back shot of narrator on cliff overlooking vast ocean, wind in his thobe", "character_type": "narrator" } }, { "scene_number": 2, "scene_category": "story", "narration_text": "Prophet Nuh, alayhi as-salam, called his people to Allah for 950 years...", "visual_direction": { "description": "Ancient city. A lone silhouetted figure stands on raised ground, arms raised to the sky, addressing a crowd below. The crowd turns away. Hot dusty afternoon.", "character_type": "story_figure", "story_element": "Nuh preaching to his people" } }, { "scene_number": 9, "scene_category": "narrator_closing", "narration_text": "And that is why patience is never wasted with Allah...", "visual_direction": { "description": "Narrator sitting on rock at sunset, calm sea, same cliff from opening but now peaceful golden light", "character_type": "narrator" } } ] } The Visual Agent uses DIFFERENT prompt strategies: narrator_opening / narrator_closing → strict narrator character prompt (brand consistency) story → story-specific prompt, unique imagery, faceless constraint on human figures only

Global Config

Before any agent runs, this config is loaded and shared with all agents. { "brand": { "channel_name": "{{CHANNEL_NAME}}", "narrator_character": "A faceless man wearing traditional white thobe and red-checkered keffiyeh/shemagh. Never show face — use back shots, silhouettes, over-shoulder angles, hands close-ups, or wide shots. Dignified, contemplative, wise.", "visual_style": "Cinematic, warm golden-hour lighting, painterly digital art style — NOT photorealistic, NOT cartoon. Epic film concept art quality.", "color_palette": ["#C8956C", "#2C1810", "#F5E6D0", "#1A3A4A", "#D4A853"], "aspect_ratio": "9:16", "resolution": "1080x1920" }, "content_guidelines": { "sensitivity_rules": [ "NEVER depict faces of any Prophet, Angel, or Sahabi", "NEVER depict faces on the narrator character", "All human figures in story scenes must also be faceless", "Use nasheeds or ambient sound only — no musical instruments", "Always include proper Islamic honorifics", "Only use Sahih or Hasan grade hadith", "Cite Surah name and verse number for all Quran references" ] }, "target_languages": ["en", "ar", "fr", "ur", "tr", "id"], "default_language": "en" }

Pipeline Execution (Orchestrator Playbook)

When the orchestrator receives a request:

Step 1: Call Story Agent

Input: { "topic": "Prophet Nuh", "target_duration_seconds": 60 } Output: Story JSON with sources, emotional arc, key visual moments

Step 2: Call Script Agent

Input: Story Agent output Output: Scene-by-scene script with: - narration text per scene - scene_category (narrator_opening / story / narrator_closing) - visual directions unique to the story - timing and subtitle text

Step 3: Call Visual Agent + Voice Agent (PARALLEL)

Visual: Script scenes → generates images per scene (narrator prompt template for opening/closing) (story-specific prompt for middle scenes) Voice: Script narration → generates audio per scene + word timestamps

Step 4: Call Assembly Agent

Input: Images + motion configs + audio + subtitle text + brand assets Output: Final MP4, thumbnail, SRT file

Step 5: Quality Checks (Orchestrator does this itself)

Verify no faces in any generated image (call face detection) Verify audio/visual sync (check durations match) Verify file size under 50MB If fail → retry failing agent up to 3 times

Step 6: Output

Save video to output directory Log to content calendar

Agent Skill Files

Each agent has its own detailed skill file: AgentSkill FilePurposeTools NeededStory Researchagents/story_agent.mdFind + validate Islamic storiesWeb search, file readScript Writeragents/script_agent.mdStory → TikTok script with scene directionsNone (pure LLM)Visual Generationagents/visual_agent.mdGenerate narrator + story scene imagesImage gen API, face detectionVoice Narrationagents/voice_agent.mdTTS with Arabic pronunciationElevenLabs/OpenAI TTSVideo Assemblyagents/assembly_agent.mdImages + audio → final videoFFmpeg/Remotion Read each agent's skill file for full input/output JSON schemas, prompt templates, and quality gates.

References

FilePurposereferences/visual_style_guide.mdFaceless techniques, environments, lighting, Ken Burns, prompt templatesconfig/global_config.jsonFull config template with API settings, brand, pipeline params

Category context

Code helpers, APIs, CLIs, browser automation, testing, and developer operations.

Source: Tencent SkillHub

Largest current source with strong distribution and engagement signals.

Package contents

Included in package

4 Docs1 Scripts1 Config

assembly_agent.md Docs
story_agent.md Docs
visual_agent.md Docs
visual_style_guide.md Docs
orchestrator.py Scripts
global_config.json Config