Useless LinkedIn AI Writer

A Chrome extension that enhances your LinkedIn content creation with AI assistance. Write better posts and comments using various AI providers, with support for images, videos, and documents.

Features

Multi-Provider AI Support

• Choose from 6 different AI providers including OpenAI, Anthropic Claude, Ollama, LM Studio, Groq, and Together AI
• Support for both cloud-based and local AI models
• Vision capabilities for analyzing images, videos, and documents

Smart Content Enhancement

• Context-aware prompts that understand whether you’re writing a post or comment
• Automatic detection of original post content when commenting
• Professional templates for various LinkedIn scenarios

Media Processing

• Extract and analyze images from LinkedIn posts
• Video frame extraction with configurable strategies
• Document carousel support for PDFs and presentations
• Configurable media processing based on your model’s capabilities

Customizable Prompts

• Create and manage your own prompt templates
• Use placeholders for dynamic content insertion
• Position-specific prompts for posts vs comments

Supported AI Providers

OpenAI

• Models: GPT-3.5-turbo, GPT-4, GPT-4-turbo, GPT-4o, GPT-4-vision-preview
• API Key Required: Yes
• Vision Support: Yes (GPT-4o, GPT-4-vision-preview)
• Endpoint: https://api.openai.com/v1/chat/completions

Anthropic Claude

• Models: Claude-3 Haiku, Claude-3 Sonnet, Claude-3 Opus, Claude-3.5 Sonnet
• API Key Required: Yes
• Vision Support: Yes (all Claude-3 models)
• Endpoint: https://api.anthropic.com/v1/messages

Ollama (Local)

• Models: Llama2, CodeLlama, Mistral, Neural Chat, Starling LM, LLaVA (7b/13b/34b), Llama3.2 Vision (3b/11b), Qwen2.5-VL (3b/7b)
• API Key Required: No
• Vision Support: Yes (LLaVA, Llama3.2 Vision, Qwen2.5-VL models)
• Endpoint: http://localhost:11434/api/generate
• Setup: Install Ollama and run ollama serve

LM Studio (Local)

• Models: Any model loaded in LM Studio
• API Key Required: No
• Vision Support: Yes (Qwen2.5-VL, LLaVA models)
• Endpoint: http://localhost:1234/v1/chat/completions
• Setup: Install LM Studio and start the local server

Groq

• Models: Llama2-70B, Mixtral-8x7B, Gemma-7B
• API Key Required: Yes
• Vision Support: No
• Endpoint: https://api.groq.com/openai/v1/chat/completions

Together AI

• Models: Llama-2-7B, Mistral-7B, RedPajama-7B
• API Key Required: Yes
• Vision Support: No
• Endpoint: https://api.together.xyz/v1/chat/completions

Installation

1. Clone this repository
2. Open Chrome and navigate to chrome://extensions/
3. Enable “Developer mode” in the top right corner
4. Click “Load unpacked” and select the extension folder
5. The extension icon will appear in your browser toolbar

Configuration

Basic Setup

1. Click the extension icon to open the configuration popup
2. Select your preferred AI provider from the dropdown
3. Enter your API key (if required for the provider)
4. Choose a specific model (optional - uses provider default if not specified)
5. Adjust temperature (0-2) and max tokens as needed
6. Click “Save Configuration”
7. Test the connection using the “Test Connection” button

Media Support Configuration

Enable/Disable Media Processing

• Check “Supports media” if your model can process images and videos
• Uncheck for text-only models to avoid API errors
• When disabled, media-related prompts are hidden from the UI

Video Frame Extraction Strategy When media support is enabled, choose how many frames to extract from videos:

• Minimal: 5 frames (start, 25%, 50%, 75%, end) - Best for small context windows
• Moderate: 11 frames (every 10% of video) - Good balance of detail and efficiency
• Dense: Every 0.5 seconds - More comprehensive coverage
• Very Dense: Every 0.25 seconds - Maximum detail (use with caution)
• Custom: Specify exact number of frames (1-100)

Choose based on your model’s context window limitations and processing capabilities.

Usage

Writing New Posts

1. Start writing a new post on LinkedIn
2. AI prompt buttons will appear below the text area
3. Click any button to enhance your content
4. The AI will consider your text and any media in the post

Commenting on Posts

1. Start writing a comment on any LinkedIn post
2. AI prompt buttons will appear with context-aware options
3. The AI will consider both the original post content and your comment
4. Media from the original post (images, videos, documents) will be analyzed if supported

Media Analysis

When commenting on posts with media:

• Images are automatically captured and analyzed
• Videos are processed using your chosen frame extraction strategy
• Document carousels provide metadata and title information
• The AI generates responses based on both text and visual content

Available Prompts

Default Prompts

Text Completion

• $ _ - Complete your text in a professional and engaging way

Professional Enhancement

• Professionalize - Convert casual text to professional LinkedIn language

Supportive Comments

• Post++ - Generate supportive comments agreeing with the post
• Comment++ - Reply to comments with agreement and additional insights

Constructive Disagreement

• Post– - Generate respectful disagreement comments
• Comment– - Disagree with comments professionally

Humorous Responses

• Post++ ;) - Funny but supportive comments
• Comment++ ;) - Humorous agreement replies
• Post– ;) - Funny but respectful disagreement comments
• Comment– ;) - Humorous disagreement replies

Translation

• ToHindi - Translate content to Hindi

Media Analysis (only shown when media support is enabled)

• Analyze Image - Analyze images and write professional comments
• Image Insights - Provide thoughtful insights based on visual content
• Video Analysis - Analyze video content and write professional comments
• Video Insights - Provide insights based on video frames
• Document Analysis - Analyze document content and write professional comments
• Document Insights - Provide insights based on document information

Custom Prompts

Create your own prompts using these placeholders:

• $text - Your current text
• $post - The original post content
• $comment - The comment you’re replying to

Local AI Setup

Ollama Setup

1. Install Ollama from https://ollama.ai/
2. Run ollama serve in terminal
3. Pull a vision model: ollama pull llava:7b or ollama pull qwen2.5-vl:7b
4. Configure the extension to use Ollama provider
5. Select the appropriate model in the extension settings

LM Studio Setup

1. Install LM Studio from https://lmstudio.ai/
2. Download and load a vision model (Qwen2.5-VL or LLaVA)
3. Start the local server in LM Studio
4. Configure the extension to use LM Studio provider
5. Select the loaded model in the extension settings

Privacy and Security

Local Providers

• Ollama and LM Studio run entirely on your machine
• No data is sent to external servers
• Complete privacy for your content

API Providers

• API keys are stored locally in Chrome’s storage
• Keys are only transmitted to the chosen provider
• No data collection or storage by the extension

Content Handling

• Your LinkedIn content is only sent to your chosen AI provider
• No intermediate storage or logging
• Media processing happens locally when possible

Troubleshooting

Connection Issues

API Key Problems

• Verify your API key is correct and has sufficient credits
• Check if the provider’s service is operational
• Ensure you have the necessary permissions for the model

Local Provider Issues

• For Ollama: Ensure ollama serve is running
• For LM Studio: Verify the local server is started
• Check that the model is properly loaded

Network Issues

• Test the connection using the “Test Connection” button
• Check your internet connection for cloud providers
• Verify firewall settings for local providers

UI Issues

No Buttons Appearing

• Refresh the LinkedIn page
• Ensure the extension is enabled in Chrome
• Check browser console for JavaScript errors
• Try disabling and re-enabling the extension

Media Processing Errors

• Uncheck “Supports media” if your model doesn’t support vision
• Try a different video frame extraction strategy
• Check if the media content is accessible

Performance Issues

Slow Responses

• Try a different model or provider
• Reduce the video frame extraction density
• Check your internet connection
• Consider using a local provider for better performance

Context Window Errors

• Reduce max tokens setting
• Use “Minimal” video frame extraction
• Switch to a model with larger context window
• Break long content into smaller pieces

Advanced Configuration

Custom Endpoints

You can override the default endpoint for any provider:

1. Enter a custom endpoint URL in the configuration
2. Useful for proxy servers or custom API deployments
3. Ensure the endpoint follows the provider’s API format

Prompt Management

Creating Custom Prompts

1. Click “Add New Prompt” in the configuration
2. Set the position (new_post or comment)
3. Enter a label and prompt text
4. Use placeholders for dynamic content
5. Choose whether to replace existing text or show in alert

Prompt Placeholders

• $text - Current text being written
• $post - Original post content (for comments)
• $comment - Comment being replied to (for replies)

Development

File Structure

├── manifest.json          # Extension manifest and permissions
├── popup.html            # Configuration UI
├── popup.js              # Configuration logic and prompt management
├── contentScript.js      # LinkedIn integration and media processing
├── contentScript.css     # Button styling and animations
├── providers.js          # Multi-provider system and API handling
├── icon.png             # Extension icon
└── README.md            # This documentation

Adding New Providers

1. Add provider configuration to providers.js
2. Define the provider’s API format and authentication
3. Add the provider to the UI dropdown in popup.html
4. Test the integration with various content types

Media Processing

The extension supports three types of media:

• Images: Direct capture and analysis
• Videos: Frame extraction with configurable strategies
• Documents: Metadata extraction from LinkedIn document carousels

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test with multiple providers and content types
5. Submit a pull request

License

This project is open source. Feel free to fork and modify as needed.

Support

For issues or questions:

• Check the troubleshooting section above
• Open an issue on GitHub with detailed information
• Include browser console logs for technical issues

---

Made by Gowrav Vishwakarma

Fork and improve this tool: GitHub Repository