Academic Social Publisher
Transform academic papers into engaging social media content for ResearchGate, LinkedIn, and WeChat. Optionally generate a video.
Input
Provide or forward the paper with:
- Title
- Authors
- Abstract / key points
- Publication venue / conference
- Link to paper
- (Optional) Specific angle or highlight to emphasize
Output Platforms
1. ResearchGate Post
- Tone: Academic, formal
- Length: 300-500 words
- Content: Full abstract highlight + methodology punchline + link
- Includes: Hashtags (#FormationControl #EventTriggered #MultiAgentSystems etc.)
2. LinkedIn Post
- Tone: Professional but conversational, storytelling-driven
- Length: 150-300 words (shorter for feed algorithm)
- Content: Hook opener → key contribution → why it matters → CTA (read paper / connect)
- Format: Short paragraphs, emoji sparingly, line breaks for readability
3. WeChat Article (公众号风格)
- Tone: Chinese, semi-formal, accessible to educated non-specialists
- Length: 800-1500 characters
- Content: 标题 (catchy) → 背景 → 核心贡献 → 亮点解读 → 结语
- Style: Use "我们" for warmth, include 2-3 emoji, technical terms with brief explanations
4. Video Introduction Script (Optional — always generated, video production depends on config)
- Length: 60-90 seconds narration (~150-200 words)
- Structure:
- Opening hook (1 sentence, intriguing question or surprising fact)
- What this paper is about (1-2 sentences)
- Why it matters (1 sentence)
- Key contribution (1-2 sentences)
- Closing CTA
- Scene/slideshow suggestions for visuals
- Output: Always produces a complete narration script
Video Production (Optional — requires config)
Video generation is opt-in and requires setup. Without an API key, the skill still produces the full script for manual use.
Option A: Full AI Video Pipeline (requires API key)
Requires one of:
- HeyGen — text-to-video with avatar, best for presenter-style research talk
- Runway / Pika / Kling — AI video clips from text/image prompts
- ElevenLabs — already configured for TTS voice
Pipeline:
- Generate narration audio via ElevenLabs (or configured TTS)
- Generate slide/image assets via image generation
- Compose video via ffmpeg
- (Optional) Enhance with HeyGen/Runway API
Option B: Script-Only Mode (no API key needed)
Even without video API keys, always:
- ✅ Generate full video narration script with scene directions
- ✅ Suggest slide content and visual layout
- ✅ Provide timing cues
User can take the script and produce the video manually or with their own tools.
Workflow
- Parse paper info (title, authors, abstract, venue)
- Identify 3 key talking points
- Draft ResearchGate version
- Adapt to LinkedIn voice
- Adapt to WeChat Chinese style
- Generate video narration script (always)
- If video API configured: produce audio/video
- Output all versions clearly separated
Configuration
To enable video generation, add to your skill config or environment:
# Example: TTS only (already has sag/ElevenLabs)
video:
enabled: true
tts: elevenlabs # uses existing sag tool
# Generate slides as images
# Example: Full pipeline
video:
enabled: true
tts: elevenlabs
video_api: heygen # or runway/pika/kling
api_key: YOUR_KEY
If video.enabled: false or no API key, skip video production and deliver script only.
Platform Best Practices
See references/platform-guides.md for detailed character limits, hashtag strategies, and posting tips per platform.